1

Addenda et corrigenda to Ranko Matasović’s Etymological Dictionary of Proto-Celtic (Brill, Leiden 2009).

Zagreb, December 2011

NOTE: This file contains only those lemmata that have been thoroughly revised, as well as about a hundred new etymologies not in the 2009 edition. Lemmata that have been subject to only minor revisions, as well as bibliographical updates, will appear in the second edition of the dictionary. Thanks to Irene Balles for helpful comments and criticisms of the first edition. *adilo- ‘target’ SEE: *ad- ‘to’ *ad-rīmo- ‘number’ [Noun] GOID: OIr. áram [ā f, later also o m] ‘act of counting’ (DIL áirem) W: MW eirif [m] ‘number, amount, sum’ SEE: *rīmā ‘number, count’ REF: GPC I: 1196, Schrijver 1995: 353. *ag-o- ‘drive’ [Vb] GOID: OIr. agid, -aig; agaid, -aga [Subj.]; acht, -acht [Pret.] W: OW hegit [3s Pres.] ‘go’, MW a, aa [3s Pres.]; aeth, ayth [Pret.] BRET: MBret. a ‘go’, OBret. nit-a; aez [Pret.] CO: Co. a ‘go’; eth, etha [Pret.] PIE: *h2e- ‘drive’ (IEW: 4f.) COGN: Skt. ájati, Lat. ago, Gr. ágō, Arm. acem ETYM: In Brit. this verb has a suppletive VN, MW mynet, MBret. monet, MCo. mones < *monetu-, cf. PCelt. *mon-ī- ‘go’. Gaul. axat (Marcellus of Bordeaux) may be 3 sg. Pres. Subj. of the same verb (Delamarre 63), so it would mean ‘he should bring’. I find it unlikely that OIr. aiged [ā f] ‘face, countenance’ should be derived from this root (the semantic connection is weak). Likewise, OIr. ág [o and u m] ‘fight, battle, valour’ might in principle be from *āgu- (or *āgo-) < PIE *h2ō-, cf. Skt. ājí- ‘race, combat’, but the long vowel (if the etymology is correct) could also be analogical after ár ‘carnage, battle’ (see *agro-). REF: LIV 255f., KPV 189-191, LEIA A-22f., DGVB 50, GPC I: 40, Delamarre 63, de Bernardo Stempel 1999: 528, Meid 1996: 45, Zair 2010: 23. *akino- ‘shooot, sprout’ [Noun] W: MW egin [p] ‘shoots, sprouts, blades’ BRET: egin PIE: *h2eḱ- ‘be sharp’ (IEW 19ff.) COGN: Lat. acus ‘needle’, OCS osъtъ ‘thistle’ ETYM: The Brittonic words have the exact cognate in Lat. acinus ‘grape or other berry’, which may point to PIE *h2eḱ-ino-. REF: GPC I: 1175. *anamī ‘blemish, fault’ [Noun] GOID: OIr. anim [?ī, later ā f] (DIL ainim) W: OW anamou gl. mendae, MW anaf BRET: MBret. anaf, anaff ‘fault’; OBret. di-anam gl. efficaciter SEE: *am-o- ‘wash’; *ono- ‘blemish’ ETYM: The PCelt. reconstruction is here based on the somewhat doubtful assumption that the OIr. word is an ancient ī-stem (it also inflects as an ā-stem, but the Nom. pl. anmi might be an indication that ī-inflection is original). Pokorny’s connection of these words with Gr. ónomai ‘blame, injure’ (IEW 779) can be retained only if one assumes (with LIV 282) that the Greek vocalism is due to assimilation (*ono- < *ano-) and that the PIE root is *h2enh3- ‘blame’ (cf. also Hitt. hannari ‘litigates,

2

sues’. In that case *anamī can also be related to *ono- ‘blemish’, cf. also MIr. antair [3 s pass.] ‘is blemished’. It is also possible (although, admittedly, speculative) that *an- is the negative prefix in this word; the root could be the same as in the verb *am-o- ‘wash’, since the meaning ‘blemish’ could have developed from something like ‘unwashed, unwashable’; cf. also the parallelism between OBret. di-anam and OIr. dianim ‘faultless’. *anawo- ‘wealth, profit’ [Noun] GOID: OIr. anae [yo m] ‘wealth, riches, prosperity’ W: MW anaw [m] ‘wealth, bounty’ GAUL: Anauo-geno, Anauus [PN], Anaone portus [Toponym] PIE: *h3enh2- ‘enjoy, use’ (IEW: 754, 47) COGN: Gr. onínēmi ‘use’, Go. ansts ‘mercy, benevolence’ ETYM: OIr. anae is mostly used in the plural. Formally, the Celtic forms could also be from PIE *h2enh1- ‘breathe’ (Lat. animus ‘spirit’, Gr. ánemos ‘breath’), but this connection is much less probable from the semantic point of view. REF: LEIA A-73, GPC I: 109, LIV 267f., Delamarre 45, 49, Zair 2010: 219. *aramo- ‘quiet’ [Adj] W: MW araf ‘quiet, leasurely, calm’ PIE: *h1erh1- ‘be still’ (IEW: 322f.) COGN: Skt. ilay- ‘be still’, rā trī- ‘night’, Av. rāman- ‘tranquility’, Gr. érēmos ‘deserted’ ETYM: If this etymology is correct (the semantic connection leaves much to be desired), PCelt. *aramo- can be regularly derived from *eramo- (by Joseph’s rule) < PIE *h1erh1-mo-. It is less probable that the root is *h1reh3- (cf. Gr. erōē ‘rest’, OHG ruowa ‘rest’, since then we would have to posit Schwebeablaut (*herh3-mo-) and Joseph’s rule to account for the Celtic forms. The connection with the Gaul. names Aramo [PN], Aramis [Hydronym], etc. is very tentative. REF: GPC I: 175., Delamarre 2003: 51, Zair 2010: 56. *argo- ‘noble, great’ [Adj] GOID: OIr. arg [o] ‘noble, great, impressive’ GAUL: Com-argus ETYM: A rather problematic etymology, since the correspondence between OIr. arg and Gaul. –argus could be accidental. Beyond Celtic, one can compare Gr. arkhós ‘leader’ and derive all of these froms from the root *h2ergh- ‘rule’. If MHG regen ‘set up, raise, stir up’ is related to these forms, we must either posit Schwebeablaut (*h2ergh- / *h2regh-), or accept that *h2rgh- was reflected as Celtic *arg- (see PCelt. *arganto- ‘silver’). REF: LEIA A-87, Delamarre 54, Zair 2010: 39. *arto- ‘bear’ [Noun] GOID: OIr. art [o m] ‘bear, hero, warrior’ W: MW arth [m and f] BRET: OBret. Ard-, Arth- , MoBret. arzh [m] GAUL: Artio [Theonym] PIE: *h2rtḱo- ‘bear’ (IEW: 845) COGN: Hitt. hartagga-, Skt. ksa-, Gr. árktos, Lat. ursus, Alb. arí ETYM: Apparently homophonous MIr. art ‘power’, ‘hero’, and ‘god’ are probably just metaphorical extensions of the word for ‘bear’ Basque hartz ‘bear’ is presumably a Celtic loanword. The development of the syllabic resonant *r > ar before a stop is probably to be explained by assuming a special development of the cluster *tḱ (to *þ or *xþ) and that this cluster behaved like the fricative *s when the syllabic resonant developed vocalic prothesis (but cf. PCelt. *mrixto-, which shows that original, non metathesised *kt was preserved in Celtic until the development of the prothesis in syllabic resonants). Thus we had *h2rtḱos > *(H)rxþos > *arxþos > *artos, but *mrktos > *mriktos > *mrixtos. Another, in my opinion less probable explanation, would involve positing a special rule *h2rC > *arC before CRC > CriC (Joseph 1982: 50f.). REF: LEIA A -91, LP 25, GPC I : 212, EIEC 55, 98, McCone 1994: 69, McCone 1996: 48, 52, Matasovic 2004: 87, Delamarre 55f., Deshayes 2003: 76, Zair 2010: 41.

3

*astili- ‘joint, part’ [Noun] GOID: MIr. asil, aisil ‘part, division, joint’ BRET: MBret. esel [m] ‘joint’, MoBret. ezel ‘limb’ CO: Co. esel ‘joint’ ETYM: Probably from the same root as in *astn(iy)o- ‘rib’, from the PIE root *h2osth1 ‘bone’ (IEW: 783). REF: LEIA A-94, Deshayes 2003: 218, Zair 2010: 65f. *ati-li-n-o- ‘stick, adhere to’ [Vb] GOID: OIr. ad-len ‘adheres to, follows’ W: MW edlynu ‘to smear, daub’ SEE: *lin-ā- ‘stick’ REF: KPV 453f., GPC I: 1166. *ati-sekwo- ‘answer’ [Noun] SEE: *sekw-o- ‘say’ *awe-C- ‘inspiration, insight’ [Noun] GOID: OIr. aui, aí [t m> f] ‘poetic inspiration’ W: MW awen [f] ‘poetic inspiration, talent’ SEE: *awelā ‘breeze, wind’ ETYM: MoBret. awen ‘inspiration’ is a loanword from W. The PCelt. reconstruction is difficult because word formations of OIr. aui and MW awen do not agree. The OIr. form points to a t-stem *h2ewh1-et-. REF: LEIA A-19, GPC I: 240, Deshayes 2003: 83, de Bernardo Stempel 1999: 172, Irslinger 2002: 57f. *axto-, *axtīno- ‘furze, gorse’ [Noun] GOID: OIr. aitten [o m] W: MW aeth, (GPC aith), eithin BRET: OBret. ethin gl. rusci CO: OCo. eythinen [Singulative] gl. ramnus PIE: *h2eḱ-sti- COGN: Lith. akstìs ‘spit (for roasting)’, Russ. ost’ ‘awn, bristle’ SEE: *akino- ‘shoot, sprout’, *akro- ‘high’ ETYM: MW aeth probably preserves the original underived noun (*axto-). The British forms presuppose a feminine noun (presumably old collective) *axtīnā, while OIr. aitten points to *axtīno- with irregular change of *xt > tt (McCone 2005a: 409 considers the possibility of a non-IE intermediary, cf. Basque ote ‘furze’). The PIE root is *h2eḱ- ‘be sharp’. REF: LEIA A-57, GPC I: 73, 1203, DGVB 168, Zimmer 2000: 521, Mallory & Adams 2006: 165. *banonā ‘young woman’ [Noun] SEE: *benā, *bena ‘woman’ *bano-wessā ‘wedding’ [Noun] GOID: OIr. banais [ā f] BRET: MBret. banues, MoBret. banvez [m] ETYM: A compound containing *bano- ‘woman’ (see *benā) and *wessā, which is derived from the same root as PCelt. *wed-o- ‘lead, bring together’. PCelt. *wessā represents PIE *wedh-teh2. The root *wedh- was used to express the concept of ‘leading the bride’ in PIE, cf. Skt. vadhū ‘bride’. REF: Irslinger 2002: 345f. *bar-na- ‘proclaim’ [Vb] W: MW barnu ‘judge, proclaim’; barn [Pres.] BRET: MBret. barn PIE: *gwerH- ‘praise’ (IEW: 478) COGN: Skt. járate ‘sings’, Lat. grātus ‘pleasing, beloved’, Lith. giriù, gìrti ‘praise’

4

ETYM: The meaning of Gaul. barnaunom (Larzac) is not quite certain; it may mean ‘judge’ or ‘judgement’, and it is probably from the same root (with the suffix *-mno-, a thematised derivative of the suffix *-mon-?). REF: GPC I: 260, KPV 213, Deshayes 2003: 94, LIV 210f., Delamarre 68. *bego- ‘break’ [Vb] GOID: OIr. do-beig, co-toibget [3p Pres.] PIE *bheg- ‘break’ (IEW 114f.) COGN: Skt. bhanákti, Arm. ebek ‘broke’, Lith. bengiù ‘end’ SEE: *bekko- ‘small’ ETYM: This verb was confused in OIr. with its near-homonym, do-boing (DIL) ‘wrests, plucks, breaks’ < PCelt. *bung-o- < *bhungo-. PIE *bheg- ‘break’ is also attested in MW di-fo, diuo ‘destroying’ (< *dī-bogo-) and in the OIr. compound ráth-buige ‘builder of ramparts’ (< *bogyo-). OIr. bocht ‘poor’ can represent *boxto- < *bhog-to- from the same root (but cf. also PCelt. *bu-n-g-o- ‘break’, and note that OIr. bocht could also be the to-participle of this root, PCelt. *buxto-). Finally, OIr. becht ‘precise, exact’ may, in principle, be from *bexto- < *bheg-to-, but the meanings are quite different so this etymology is improbable in my opinion. REF: KPV 216f, McCone 1991, LIV 66f., Irslinger 2002: 248f., 262. *belyo- ‘tree’ [Noun] GOID: MIr. bile [io n] ‘large tree, tree trunk’ PIE: *bholh3yo- ‘leaf’ (IEW: 122) COGN: Lat. folium, Gr. phýllon SEE: *blātu- ‘flower’ ETYM: The fact that the vocalism of OIr. does not agree with the (expected) o-grade in Gr. and Lat. presumably shows that these are parallel formations from the root *bhelh3-. The laryngeal probably dropped before *y in PIE (Pinault’s rule). In Gaulish, the same root is probably found in toponyms such as Billio-magus (> Fr. Billom). Pr. bilha ‘tree-trunk’ may be from the (unattested) Gaulish reflex of this word. REF: LEIA B-50f., Delamarre 75, Zair 2010: 213. *berantī ‘sow’ [Noun] SEE: *ber-o- ‘carry, bear, bring’ *bēto-, *biyato- ‘food’ [Noun] GOID: OIr. bíad [o n] W: MW bwyd [m] BRET: MBret. boet, MoBret. boued [m] CO: OCo. buit gl. cibus vel esca PIE: *gweyh3- ‘live’ (IEW: 468) COGN: Lat. uīta ‘life’, Lith. gyvatà SEE: *biwo- ‘alive’ ETYM: OIr. bíad is attested as a bisyllable in early sources, which implies that the proto-form was *biwoto- < *gwih3-wo-to- (cf. Lat. uīta < *wiwotā, Lith. gyvatà), or, slightly more probably, *biyato- < *gwih3eto- (cf. Gr. bíotos ‘life’), with the suffix *-to- added to the zero-grade of the root (cf. PCelt. *riyatro- ‘torrent’ < *riHetro-); the Brittonic words can be derived from *bēto-, ultimately form PIE *gweyh3-to-. A derivative is found in OBret. boitolion gl. esciferis. REF: LEIA B-47f., GPC I: 358, 215f., Deshayes 2003: 124, Hamp 1976, Irslinger 2002: 249f., Zair 2010: 270. *bi-na- ‘strike, hit’ [Vb] GOID: OIr. benaid; -bia [Subj.]; bíu [1s Fut.]; bí [Pret.]; -bíth [Pret. Pass.] W: MW kymynu ‘hit, cut down’; kymyn [3s Pres.] BRET: MBret. benaff ‘cut’ GAUL: biietutu ‘should strike’ [Impv.] (Larzac) CELTIB: ne-bintor ‘should not be hit’ [3p Impv. Med.] (Botorrita I) PIE: *bheyH- ‘strike’ (IEW: 117f.) COGN: OLat. perfines ‘you should strike’, OCS biti ‘strike’

5

SEE: *biyatli- ‘axe’ ETYM: MW kymynu is from *kom-bi-na- (the simplex verb is unattested in Welsh). The interpretation of Celtib. nebintor is dubious, but it appears probable that it is a verbal form with the middle ending, and the etymological connection with the root *bheyH- is accepted by most specialists. OIr. bíth ‘blow, wound’ (also as a preposition ‘because’) is from the same root (PCelt. *bītV-), but the stem formation of this old verbal noun of benaid is unknown. I doubt that MW bid [f] ‘bush, thorn’ is from the same source, since the meanings are too different. REF: LIV 72, KPV 226ff., LEIA B-32-34, McCone 1991: 11, 125, RIG II.2: 251-266, Lambert 1994: 160-172, 214f., MLH V.1: 271f., Irslinger 2002: 431. *biyatli- ‘axe’ [Noun] GOID: OIr. biáil, biail [i m] ‘axe, hatchet, battle axe’ W: OW bahell gl. securis, MW bwell, buyall (GPC bwyall [f] ‘axe’ BRET: MBret. bouhazl, bouchazl, MoBret. bouc’hal [f] ‘axe’ CO: MCo. boell ‘axe’ SEE: *bi-na- ‘strike’ ETYM: Under the hypothesis that *H was lost after *ey in Celtic, OIr. biail cannot be from *beyatli- < *bheyH-tli-; rather, PCelt. *biyatli has generalized the shape of the root *bi- from the verb *bi-na-. Since *biyatli- is almost certainly derived from earlier *biya-tlo- (with the suffix *-tlo- used in formation for nouns denoting instruments), it is also possible that the shape of the suffix *-atlo- is analogically introduced from words such as *banatlo- ‘broom-plant, broom’, where *-a- is the regular reflex of *H between consonants. Note that the source of –h- in MBret. bouhazl is unknown, and the relationship of W bwyall and MCo. boell to these forms is unclear, since *tl would have been preserved in Welsh and Cornish as *dl. REF: LEIA B-47, McCone 1996: 123, Zair 2010: 282. *blāro- ‘grey’ [Adj] SEE: *blāwo- ‘yellow’ *blātu- ‘flower’ [Noun] GOID: OIr. bláth [u m] W: MW blawd [m] BRET: MBret. bleuzff, MoBret. bleuñv [Collective] ‘flowers’ CO: OCo. blodon Co. bledzhan PIE: *bhleh3- ‘blossom, flower’ (IEW: 122) COGN: Lat. flōs, OHG bluot ETYM: Gaul. PN Blatuna is probably derived from the same root (Meid 2005: 187), cf. also the toponym Blato-magus. PCelt. *blātu- presupposes PIE *bhleh3-tu-, an abstract nominal derivative (’blossoming’), but it is also compatible with *bhlh3-tú-, if one assumes that the root vowel was shortened by Dybo’s law. The zero-grade of the root is attested in OHG blat ‘leaf’, ToA pält ‘leaf’... REF: LEIA B-58, GPC I: 384, Deshayes 2003: 114, de Bernardo Stempel 1999: 289, Irslinger 2002: 86, Meid 2005: 187, Sims-Williams 2006: 45. *borro- ‘big, strong, stiff’ [Adj] GOID: MIr. borr [o] ‘thick, big, swollen’ W: MW bwrr ‘thick, big’ CO: OCo. bor gl. pinguis PIE: *bhers- ‘point’ (IEW: 109) COGN: Skt. bhṛsṭí- ‘tip, point’, OHG borst, OIc. barr ‘sharp’ SEE: *barro- ‘point, top’ ETYM: MIr. buirre ‘swelling’ is a derivative (PCelt. *borryā); PCelt. *borro- probably represents the o-grade of the root attested in *barro- ‘point, tip’. REF: LEIA B-73, Hill 2003: 272, Schrijver 1995: 55, 65ff. *bostā ‘palm, fist’ [Noun] GOID: OIr. bos, bas [ā f] W: OW bos, MW bos [f] BRET: MBret. boz

6

PIE: *gwos-to-, *gwos-dho- ‘branch’ (IEW: 382, 480) COGN: MHG quast ‘branch’, Alb. gjethe ‘leaf, foliage’ SEE: *buzdo- ‘penis’ ETYM: Fr. boisse ‘measure of grain, bushel’ can be derived from Gaul. *bostiā (Gamillscheg 124). Romance words denoting ‘quantity that can be taken by two hands, two handfuls’ such as Catalan almosta, Piémontais ambosta and OSp. ambuesta seem to be Gaulish loanwords (from PCelt. *ambi-bostā). Some linguists derive these Celtic words from non-IE sources, comparing them, e.g., to Basque bost ‘five’ (allegedly from ‘the number of fingers on a palm’), but these are mere speculations, and a good IE etymology is available. If the Basque word is related at all, it could have been borrowed from Celtiberian. The Proto-Celtic form presupposes PIE *gwos-to-, which should be a derivative from the same root as *gwos-dho-, with a different suffix (cf. OCS gvozdъ ‘nail’), from which it may be possible to derive PCelt. *buzdo- ‘penis’. REF: LEIA B-20f., GPC I: 302, Falileyev 17, Delamarre 42, Gamillscheg 124, Irslinger 2002: 372f. *bow- ‘cow’ [Noun] GOID: OIr. bó [irregular f] W: MW bu, buw [m and f] BRET: MBret. bou-tig ‘stable’ GAUL: Bo-marus [PN] CELTIB: bou-stom (?) ‘stable’ (Botorrita I) PIE: *gwōw- ‘cow’ (IEW: 482) COGN: Lat. bōs, OHG chuo, Skt. gáu-, ToA ko, Arm. kov ETYM: The exact interpretation of Celtib. bou-stom is uncertain, but it is probable that it contains the root *bow- ‘cow’. It might be formally identical to OIr. búas [o ?n] ‘riches, wealth (in kine)’ < *bow-sto-. The MW form bu is used only in counting cows; otherwise, the derivative buch is used, with parallels in MBret. buch, Co. buch (from *bow-sko-). The attested forms allow the reconstruction of a root-noun in PCelt., with Nom. sg. *bāws, Gen. sg. *bow-os, Acc. sg. *bow-am (instead of PIE *gwōm which would yield PCelt. **bām), etc. W biw ‘horned cattle’ can be derived from the acc. pl. of the word for ‘cow’ (PIE *gwōns > PCelt. *būs), with the new Proto-British form on the analogy with the consonant stem *būwās, which yields W biw regularly (Isaac 2007: 34). This is clearly preferable than deriving W biw from the PIE adjective *gwih3wó- ‘alive’ (see PCelt. *biwo-), where we would expect Dybo’s law to operate. REF: LEIA B-61, GPC I: 342, DGVB 88, LHEB 641, EIEC 98, 134, Delamarre 79f., Jordán Cólera 1998: 14, 39, Villar 1997: 907, MLH V.1: 87, Isaac 2007: 34, Wodtko et alii 2008: 189ff.. *bowd-ro- ‘dirty’ [Adj] GOID: MIr. buaidir [?i] ‘confusion’ W: MW budyr ‘filthy, dirty, mean’ (GPC budr) PIE: *gwew-dh- ‘excrement, defecate’ (IEW: 484) COGN: Skt. guváti ‘cacat’, OE cwéad ‘dirt’ ETYM: The meaning of MIr. búaidir is not completely clear (it is compatible with the meaning ‘confusion’). Cf. also OIr. búaidre [iā f] ‘trouble, confusion’. The meaning ‘confusion’ could have developed from something like ‘mess’ and ‘filth’ which may be posited for Brittonic. Cf. also MW baw [m] ‘dirt, filth’ and Fr. boue ‘mud’ (from Gaulish) which might be related. The etymology offered here is in any case very tentative. REF: LEIA B-108, GPC I: 344, Zair 2010: 269f. *brig- ‘hill’ [Noun] GOID: MIr. brí [g f] W: MW bre [f] BRET: MBret. bre [m] CO: Co. bre GAUL: -briga [in Toponyms], Brig-antes [Ethnonym] PIE: *bherh- ‘be high, hill’ (IEW: 140f.) COGN: Av. bərəz-, OHG berg SEE: *brigā ‘might, power’ ETYM: The Celtic forms can be derived from the zero-grade of the PIE root *bherh-, and the observed vowel alternations point to a PIE root noun (Nom. sg. *bhērhs, Gen. sg. *bhrhos, Acc. sg.

7

*bherhm). Celtic generalized the stem of the oblique cases, hence the paradigm PCelt. Nom. sg. *brig-s, Gen. *brig-os. The root noun is preserved in OIr. brí, while the Brit. and Gaul. forms point to a derivative *brigā. OIr. Brigid ‘dea poetarum’ (Cormac), OW brennhin, brennin, MW brenhin ‘king’ are from the same root; they presuppose *brigantī (= Skt. bhatī- ‘the high one’, an epithet of Uṣas), and *brigantīnos, respectively. The original meaning would be ‘the exalted one’. The latter word is attested in Gaul. as birikantin on a coin sometimes falsely attributed to Celtiberian, because it is written in Iberian script. However, it is discovered in Southern France, so it is presumably Gaulish (MLH V.1: XII). REF: LEIA B-87, GPC I: 313, LP 30, LIV 78f., EIEC 269, Lambert 1994: 20, 37, Delamarre 87, Deshayes 2003: 133, Sims-Williams 2006: 50ff. *brīgo/ā ‘might, power’ [Noun] GOID: OIr. bríg [ā f] W: MW bri [m] ‘prestige, authority’ BRET: MBret. bry ‘respect’ CO: Co. bry ‘respect, honor’ SEE: *brig- ‘hill’ ETYM: Note that the genders of W and Goidelic do not match. The connection with *brig- ‘hill’ is improbable, because of the long *-ī- in this word. The etymology relating PCelt. *brīgV- and Gr. bri thos ‘weight’, Latv. grins ‘angry < *gwriH-g- (IEW 477) is very tentative and semantically quite implausible. REF: LEIA B-90, GPC I: 323, de Bernardo Stempel 1999: 30, Zair 2010: 145. *bristi- ‘haste’ [Noun] W: MW brys [m] ‘hurry, haste’ BRET: MBret. bresic, brezec ‘swift’ PIE: *bhris-ti- ‘haste’ COGN: Lat. festīno ‘hurry’ ETYM: The Breton form is formed with the suffix *-iko-. In Gaulish, one may compare the PN Brista (Limoges, attested in Gen. sg. Bristas). Latin festīnāre is built on an unattested ti-stem *festis (Schrijver 1990), which, just as PCelt. *bristis, must come from PIE *bhris-tis. A deeper connection with the PIE root *bhreyH- ‘cut’ (OCS briti, etc., LIV 92f.) appears difficult to me, both formally (there are no traces of the laryngeal in *bristi-) and semantically. REF: GPC I: 340, Delamarre 88, Schrijver 1990, Irslinger 2002: 199, de Vaan 2008: 216. *brixtu- ‘magical formula, incantation’ [Noun] GOID: OIr. bricht [u n] W: MW -brith in lled-frith ‘magical charm’ BRET: OBret. brith GAUL: brixtia (Chamalières) PIE: *bherh- ‘enlighten (?)’ COGN: ON bragr ‘poetic talent’, Skt. brahmán- ‘priest’ SEE: *berxto- ‘bright, beautiful’ ETYM: The PIE root must be reconstructed without the laryngeal, otherwise we would expect *CrHC > CrāC, PCelt. **brāxtu-). Therefore, it is doubtful whether this is the same root as *bherH- ‘bright’ (IEW 139, cf. Skt. bhrā jate ‘shines’, Go. bairhts ‘bright’, etc.); we may want to compare PCelt. *brixtu- with *berxto- ‘bright, beautiful’, which also has to be from a root without a laryngeal. The same root may be found in MIr. brigaid ‘shows, declares’, MW bryth-ron ‘magic wand’ and Gaul. brixtom ‘magical formula’ (Larzac). REF: LEIA B-89, Delamarre 90, Fleuriot 1976-7: 181, Lambert 1994: 154, Irslinger 2002: 87f., de Vaan 2002. *brokko- ‘badger’ [Noun] GOID: MIr. brocc [o m], Ogam BROCI W: MW broch [m] BRET: MoBret. broc’h [m] CO: OCo. broch gl. taxo

8

GAUL: Broco-magus [PN] ETYM: This word has no known cognates in other IE languages, and it has all but replaced the inherited word for ‘badger’, *tasko-. It may have been borrowed from some non-IE language into Proto-Celtic, cf. also MoIr. broc ‘grey, speckled’, brocach ‘dirty, filthy, spotted, grey, clumsy’ and W broc ‘of a mixed colour’ (a loanword from Irish), which are probably related and may indicate that ‘badger’ was originally ‘the grey one’. Balles (2010) derives PCelt. *brokko- from PIE *bhrog-ko- (with the same suffix as in *bukko- ‘goat’ and *mukko- ‘pig’ (here reconstructed as *mokku-). She identifies the root with that of *brag-yo- ‘fart’, noting that badgers smell badly like all Mustelidae. However, this etymology is difficult because PCelt. *brag-yo- is from a PIE root in laryngeal, so we would expect PCelt. *brāk- from PIE *bhroHg-. REF: LEIA B-94, GPC I: 330, LHEB 567, Lambert 1994: 191, Delamarre 90, Campanile 1974: 18, Deshayes 2003: 140, Ziegler 1994: 100, Sims-Williams 2006: 56. *brond-ī- ‘spring forth, rush’ [Vb] SEE: *brend-o- ‘flow, spurt out’ *browso- ‘belly’ [Noun] SEE: *bruson- abdomen, womb’ *brū- ‘brow’ [Noun] GOID: OIr.-brú (for-brú) [u f] PIE: *(h3)bhrewH- ‘brow’ (IEW: 172f.) COGN: Skt. bhrū-, Gr. ophrýs, OE brú, Lith. bruvìs, Russ. brov' ETYM: PCelt. *brū- is certainly related to PCelt. *abrant- ‘eyelid’ (OIr. abrae [nt m], MW amrant, MoBret. abrant, Co. abrans), but the exact nature of the relationship is unclear. Hamp (1981: 49ff.) posits PIE *h1p-bhrnt- to account for a- in *abrant-, where *h1p- would be the zero grade of the root found in the preposition *h1epi- (Gr. epí ‘on’). This is formally possible, but quite speculative. A connection may also exist with the word for ‘bridge’, PCelt. *brīwā. The word-initial *h3- in the PIE reconstruction can be posited only on the basis of Gr. o- in ophrýs. REF: LEIA B-75, EIEC 188, Wodtko et alii 2008: 41ff., Zair 2010: 64f. *brugno- ‘sadness, pain’ [Noun] GOID: OIr. brón [o m] W: MW brwyn [m] ETYM: The derivation of this Celtic word from PIE *gwrewHh- ‘bite’ (IEW 486) is difficult, as Gr. brkhō, Lith. gráužiu, OCS gryzetъ ‘bites’ etc. clearly point to a laryngeal in the root, of which there is no trace in Celtic. Perhaps we should assume PIE *gwruHh-nó- and subsequent shortening of the first vowel by Dybo’s law, or the laryngeal was lost before a consonant cluster (the so-called ‘Wetter-rule’, which is not generally accepted). Note, however, that W brwyn is also compatible with PCelt. *brugino- and *brogino-, and OIr. brón can be from *brogno- (cf. OIr. srón ‘nose’ < *srognā). Thus, one may want to connect PCelt. *brogno-, *brogino- to PIE *bhreg- ‘break’ (Lat. frango, Go. brikan). REF: LEIA B-96, GPC I: 336, Zair 2010: 200. *bruson- ‘abdomen, womb’ [Noun] GOID: OIr. brú [n f] W: MW bron [f] ‘breast’ BRET: OBret. bronn ‘breast’, MBret. and MoBret. bronn ‘breast, teat’ [f] CO: Co. bron ‘breast’ PIE: *bhrews- ‘belly’ (?)(IEW: 170f.) COGN: OE breost ‘breast’, Russ. brjúxo ‘belly, paunch’ PCelt. *bruson- is an n-stem derived from the zero grade of the PIE root (PIE *bhrus-on-). The nominative sg. was *brusū (> OIr. brú), and the genitive singular *brusnos (> OIr. bronn). MW bron and the Breton and Cornish forms point to *brusnā, which may be formed from the oblique cases of the original n-stem. MW bru [m] ‘womb, belly’ is from a different ablaut grade of the root, PCelt. *browso- < PIE *bhrewso-, with cognates in Slavic, cf. ORuss. brjuxo ‘belly’, Pol. brzuch ‘id.’ The Gaulish reflexes of this word are preserved in Fr. dial. bronne ‘teat’ (Maine), abron ‘teat’ (Berry), and perhaps in toponyms such as Bronium, Broniacus, etc.

9

REF: LEIA B-100, GPC I: 334, DGVB 90, EIEC 561, Delamarre 92, Deshayes 2003: 142, de Bernardo Stempel 1999: 100, Stüber 1998: 112f., Derksen 2008: 63. *bruto- ‘prick’ [Noun] GOID: MIr. broth [o] ‘beard, ear of corn, sedge (?)’ BRET: MoBret. broud ‘nail’ CO: Co. bros gl. aculeus COGN: Lat. frutex ‘shrub, bush’ ETYM: The connection of the Celtic words and Lat. frutex is somewhat doubtful. If they are indeed related, the PIE root would be *bhrewt-. The basic meaning is ‘something sharp’. REF: LEIA B-98, Deshayes 2003: 142, de Vaan 2008: 245, Irslinger 2002: 262f., Zair 2010: 177. *budaro- ‘deaf’ [Adj] GOID: OIr. bodar [o] W: MW byddar ‘deaf, numb’ BRET: OBret. bodaran, MBret. bouzar CO: OCo. bothar gl. surdus, Co. bodhar PIE: *bhod(h)Hro- ‘deaf’ (IEW: 112) COGN: Skt. badhirá- ‘deaf’ ETYM: The Brittonic reflexes show that the PCelt. form was *budaro-, not *bodaro-. The comparison with Skt. badhirá- is possible only under the assumption that the vocalism in Celtic is analogical, perhaps under the influence of (unattested) Celtic reflexes of the root *bhowdh- > Go. bauþs ‘deaf’. Of course, this is very speculative. The Gaulish PN Bodaro may be from the same root with the original vocalism. REF: LEIA B-64f., GPC I: 362, DGVB 87, EIEC 148., Campanile 1974: 16, Delamarre 80f., Schrijver 1995: 52, Zair 2010: 253.. *bu-yo- ‘be, become’ [Vb] GOID: OIr. biid ‘is wont to be’ [Consuetudinal Pres.]; -bé [Subj.]; bieid, -bia [Fut.]; boí [Pret.]; -both [Pret. Pass.] W: OW bot (GPC bod); bez [3s Pres.]; bit [3s Pres.]; bei [Impf. Subj.]; bu [3s Pret.] BRET: OBret. bout; bei [3s Impf. Subj.]; boe, boue [3s Pret.] CO: Co. bos; beth [3s Pres.]; be [3s Impf. Subj.]; bue [3s Pret.] GAUL: biiete [2p Impv] (Inscription from Limé) PIE: *bhewh2- ‘be, become’ (IEW: 146-150) COGN: Skt. bhávati, Lat. fuī ‘I was’, Lith. bū ti, OCS byti SEE: *es- ‘be’ ETYM: These forms are suppletive to *es- ‘be’; OIr. biid (disyllabic) ‘is wont to be’ and Gaul. biiete point to PCelt. *bīye- < *bhwiye- (< *bhuh2i(H)ye-), cf. Lat. fio ‘become, be’ (Kortlandt 2007: 136). There is considerable disagreement about the reconstruction of the paradigm of the OIr. and W preterite; OIr. 3 sg. boí and W bu can be from PCelt. *bowe < PIE *bhowe (perfect without reduplication), but OIr. 1sg. -bá is a problem. Kortlandt (2007: 125) now reconstructs the PIE root as *bheh3u- and derives -bá from *bōum < *bheh3um. The verbal noun of the verb ‘to be’ in OIr. is buith, both < *butā, with parallels in Brittonic, cf. W bod [m] ‘being, existence’, OBret. bot ‘being’. The masculine gender in W is secondary, since the original *u was lowered to o. The short root vowel may be due to Dybo’s law. REF: KPV 241-256, GPC I: 292, LP 325-331, LEIA B-46f, McCone 1991: 115-135, RIG II.2: 269-280, Morris Jones 1913: 346-353, Kortlandt 2000, 2007: 136, Irslinger 2002: 400f., Schumacher 2000: 48, 66ff., Wodtko et alii 2008: 46ff. *dā- ‘give’ [Vb] GOID: OIr. -tartat ‘could give’ [3s Pres.] do-rat ‘has given’ [Perf.] GAUL: dede [Pret.] (Orgon, etc.) LEP: tetu (Prestino) CELTIB: tatuz [Imp.] (?) (Botorrita I) PIE: *deh3- ‘give’ (IEW: 223ff.) COGN: Skt. dā-, Lat. do, dare, Gr. dídōmi, Lith. dúoti, OCS dati

10

SEE: *dānu- ‘gift’ ETYM: OIr. tartat < *tu-fro-ad-dā-. On Celtib. tatuz (Botorrita I) see Eska 1989: 142); some derive this form from the root *dheh1- ‘do’ (Lat. facio, Gr. títhēmi, etc.); the root *deh3- is probably also attested in Celtib. zizonti ‘they give’ (by assimilation from *dizonti) and taunei ‘to give’. Gaulish dede and Lepontic tetu might also be from PIE *dheh1- rather than from *deh3-, but this appears less probable. OIr. dúas [ā f] ‘gift, reward, recompense to poets’ is derivable from PCelt. *dowstā; it might be from PIE *deh3- with the suffix *-w- (perhaps originally the labial feature of the laryngeal *h3?) as in Lat. duim ‘would give’, Lith. da vė ‘gave’. Gaul. readdas (Saint-Marcel near Argenton-sur-Creuse) has been analysed as *fr(o)-e-ad-dā-s ‘he has given this’ (Schrijver 1997: 178f.), but this is just a possibility. Finally, OIr. íadaid, -íada ‘closes’ is certainly an old compound with the prefix *efi- (< PIE *h1epi-, cf. Gr. epí), but it is unclear whether the verbal root is *deh3- ‘give’ (as assumed by Schumacher in KPV), or *dheh1- ‘do, make, put’. The latter appears more probable from the semantic point of view. REF: KPV 265 ff., LIV 105, Delamarre 138f., Lambert 1994: 137f., Lejeune 1971: 94, Prosdocimi 1984: 73, MLH V.1: 365ff., Corthals 1979. *damniyo- ‘material, matter’ [Noun] GOID: OIr. damnae [yo m] W: MW defnydd [m] ‘matter, use, cause’ BRET: MBret. daffnez, MoBret. danvez PIE: *demH- ‘build’ (IEW: 198f.) COGN: Gr. démō, Go. ga-timan ‘fit’ ETYM: Underived *damno- may be preserved in W dafn [m] ‘yarn on weaver’s loom, warp and woof’, but the semantic connection is weak. The PIE root is reconstructed as *demh2- by LIV because of Gr. démas ‘(man’s) body’. REF: LEIA D-21, GPC I: 913, LIV 114f. *dawno- ‘drop’ [Noun] W: W dafn [m] COGN: OE dēaw ‘dew’, Germ. Tau ETYM: W dafn is derivable from *dheHu-no-, while the Germanic forms represent *dheHu-wo-. In principle, these words could be from the same root as in Lat. fūmus ‘smoke’, OCS dymъ < PIE *dhuh2mo- (IEW 261), but both the semantic and the formal connections are somewhat doubtful. REF: GPC I: 878, Kluge 817. *dexs(i)wo- ‘right, south’ [Adj] GOID: OIr. dess [o] W: MW dehau, deheu (GPC deau, de) BRET: OBret. dehou, MBret. deho, dechou, MoBret. dehou, dehow CO: Co. dehow, dyghow GAUL: Dex(s)iua [Theonym] PIE: *deḱs-wo- ‘right’ (IEW: 190) COGN: Lat. dexter, Gr. deksiterós, OCS desnъ, Alb. djathtë ETYM: The Insular Celtic forms are derivable from *dexswo-, while the Gaulish name Dexsiua points to *dexsiwo-, which is the expected form, in light of the cognates in other IE languages (cf. also Gr. deksiós < *deḱsiwo-). The meaning ‘south’ is derived from the main principle of orientation in the Celtic and IE tradition, by facing the rising sun (cf. also OIr. túath ‘left, north’ < PCelt. *towto-. Another derivative from the same PIE root is MW destl ‘orderly, tidy’, probably from *dexstlo-. W des [m] ‘system’ (hapax) and MIr. des ‘arrangement, order’ are poorly attested, but, in principle, they could be from the same root. REF: Stokes 145, GPC I: 934, 999, DGVB 133, LEIA D-60, 61f., EIEC 458, LHEB 535, Deshayes 2003: 174, de Bernardo Stempel 1999: 212, Delamarre 143, Beekes 1994. *dīko- ‘anger’ [Noun] W: MW dic (GPC dig) [m] ‘anger, wrath, grief’ PIE: *diHk- (IEW 187) COGN: Russ. díkij ‘wild’, Pol. dziki ‘wild’, Latv. dīks ‘empty, idle’, Lith. dy kas ‘empty, idle, vacant’

11

ETYM: A very tentative etymology, since the evidence is limited to Welsh and Balto-Slavic. Derksen (2008: 107) thinks that the Baltic words might be borrowed from Slavic because they have circumflex on the root, while the Slavic cognates point to the acute from PIE laryngeal. He considers it possible that the Slavic words are cognate with Skt. day- ‘fly (of birds, chariots, gods)’ but does not mention W dig. REF: GPC I: 994, Zair 2010: 147. *di-na- ‘suck’ [Vb] GOID: OIr. denaid, denait [3p Pres.]; did [Pret.] W: MW dynu, denu BRET: MBret. denaff, MoBret. denañ CO: Co. dena PIE: *dheh1y- ‘suck’ (IEW: 241f.) COGN: Luv. titaimi- ‘fed’, Skt. dháyati, Lat. fēlāre, Go. daddjan, OCS dojiti ‘suckle’, Arm. diem SEE: *dilā ‘teat’ ETYM: The PCelt. present stem *di-na- cannot be derived from *dhi-neh1- (this would have given OIr. **dinid rather than denaid), so the suffix *-na- was extended analogically after the large class of Celtic presents with that suffix. The root *dhi- is the result of laryngeal metathesis (*dh1i-C- > *dhih1C-). Derivatives from this root include OIr. dínu [nt m] ‘lamb’, W dyniawed, dynawed, dyniewed [m] ‘yearling, stirk, young bullock’, OCo. denevoit gl. iuvencus, but a common PCelt. form cannot be reconstructed on the basis of these words. MIr. did ‘teat’ is attested only twice; in principle, it could be from the same root, but the stem formation and gender are unclear, as well as the reason why the vowel –i- is short. REF: KPV 273ff., LEIA D-49, GPC I: 1140, LP 36, EIEC 556, LIV 138f., Deshayes 2003: 178, McCone 1991: 14f., Zair 2010: 151. *dī-reyo- ‘fine, honour-price’ [Noun] GOID: OIr. díre [yo n] ‘honour-price, penalty, mulct’ W: MW dirwy [m and f] ‘fine, mulct, penalty’ ETYM: This word is composed of the preposition *dī- and the stem *reyo-, which is unattested as simplex. It may be from the same root as in *rīmā ‘number, count’, *ri-na- ‘count, sell, exchange’. REF: LEIA D-95, GPC I: 1038, Zair 2010: 300. *dowstā ‘gift’ [Noun] SEE: *dā- ‘give’ *dowyo- ‘slow’ [Adj] GOID: OIr. dóe [io] gl. tardus ‘slow, sluggish’ PIE: *duh2- ‘far, long distance’ (IEW: 219f.) COGN: Skt. dávīyas- ‘further, more distant’, Gr. dē n ‘far’ (Adv.), Lat. dū-dum ‘some time ago’, Arm. tev ‘duration’, perhaps Hitt. tūwaz ‘from afar’ ETYM: PCelt. *dowyo- < PIE *dowh2-yo- with the o-grade of the root. The loss of the laryngeal is regular. In PIE, an ablauting paradigm must be assumed: Nom. sg. *dwōh2-s, Acc. sg. *dweh2-m (cf. Gr. dē n < *dwān), Gen. sg. *duh2-os (cf. Lat. dū-dum). REF: LEIA D-204, de Bernardo Stempel 1999: 204, Schrijver 1991: 232, Zair 2010: 214. *dri-n-g-o- ‘climb, advance’ [Vb] GOID: OIr. dringid, -dring; dreisi [2s Subj.]; dreblaing [Pret.] W: MW dringo (GPC dringo, dringio, dringad) PIE: *dregh- ‘hold’ (IEW: 212f., 254) COGN: Av. dražaite ‘holds’, Gr. drássomai, Russ. deržát’ ‘hold’ ETYM: OIr. drécht [u m] ‘part, portion’ may be derivable from *dreng-tu- and preserve the original meaning of the verb ‘hold’ implied by the cognates in other IE languages. In that case, the nasal in *dreng-tu- must be based on the analogy with the present stem. MIr. dréimm ‘climbing’ is the verbal noun to OIr. dringid and probably goes back to *drinxsman- (with *-n- from the present stem). REF: KPV 285f., GPC I: 1085, DGVB 152, LIV 126, LP 359, LEIA D-196, McCone 1991: 47, Stüber 1998: 70, Irslinger 2002: 142f.

12

*drixsmā ‘face’ [Noun] SEE: *drikā ‘face, appearance’ *drixsnā ‘quarrel’ [Noun] GOID: MIr. drenn [ā f?] ‘quarrel, combat’ BRET: OBret. ardren gl. praepugnis (DGVB ardrén) PIE: *dhregh- ‘incite, provoke’ (IEW 273) COGN: OCS raz-dražiti ‘incite (against), provoke’, Croat. drážiti ‘annoy’ ETYM: A very tentative etymology. MIr. drenn cannot be related to *der(H)- ‘tear’ (see *darno- ‘piece, part’), since we would expect **darn- from both *drno- and *dr-sno-. The accentuation of Croat. drážiti points to original *ō in the root (if there were a root-final laryngeal, we would have short a). Hence, the Slavic forms are probably denominative, going back to an (unattested) root noun *dhrōgh-s ‘quarrel’. If MIr. drenn is indeed related, it must be from the zero-grade form of this root (*dhrgh-) with the common suffix *-sno-. OBret. ardren presupposes PCelt. *fare-drixsnV-, but it could also be from *fare-trexsnV-, cf. PCelt. *trexsno- ‘strong’. Delamarre (328) compares also the Gaul. ethnonym Uoto-drones, but this is just a conjecture. REF: LEIA D-193, Delamarre 328, DGVB 72. *drūxtu- ‘drop’ [Noun] GOID: OIr. drúcht [u m] ‘drop (of dew)’ PIE: *dhrewb- ‘drop’ (IEW 275) COGN: OHG troffo, tropfo ‘drop’ ETYM: This etymology is quite uncertain, as the long ú in OIr. is unexpected. The length is preserved in MoIr. drúcht ‘dew’. If the etymology is correct, the only cognates are found in Germanic, cf. also OHG triufan, OIc. drjúpa ‘drop’, with the e-grade of the root. The alternation in vowel length may indicate that the root is borrowed from some non-IE source. REF: LEIA D-202, GOI 140, LIV 156, de Bernardo Stempel 1999: 526, Irslinger 2002: 96. *dwīyot- ‘smoke’ [Noun] GOID: OIr. dé [d f]; díad [Gen. s] PIE: *dhewh2- ‘make smoke, fumigate’ (IEW: 261ff.) COGN: Lat. suffio ‘fumigate’, Lith. dial. dujà ‘mist’, ToB tweye ‘dust’ ETYM: Cf. also OIr. dethach [ā f] ‘smoke’ < *dwit-ākā. W dew ‘fog’ is a ghost-word. We should probably start from PIE *dhuh2-yo-, which developed as *dhwīyo-, the immediate proto-form of Lat. suf-fio (Kortlandt 2007: 136). In Celtic, a dental suffix was added to this stem, hence PCelt.*dwīyot-. REF: LEIA D-28, Kortlandt 2007: 136, Irslinger 2002: 61. *exs ‘out of, from’ [Prep] GOID: OIr. ess-, ass-, a W: MW ech, eh BRET: OBret. ech GAUL: ex- CELTIB: es, es- PIE: *(h1)ehs (IEW: 292f.) COGN: Lat. ex, Gr. eks, Lith. iš-, OCS iz ETYM: W eithaf ‘extreme, farthest’ (also nominalized as ‘extremity, end’) is derivable from *extamo-, with the superlative suffix *-tamo- added to the root of *exs. REF: GPC I: 1160, 1202, DGVB 154, GOI 507ff., Delamarre 169, Ó Flaithearta 1997: 658, MLH V.1: 128. *fanssā ‘footprint’ [Noun] GOID: OIr. és [ā f] (DIL éis) ‘track, trace’ PIE: *peth2- ‘spread out’ (IEW: 824f.) COGN: Lat. pando ‘spread out, extend’, pateo ‘be open’, passum ‘step’, Gr. pítnēmi ‘spread out, open’ SEE: *fatamā ‘palm of the hand’ ETYM: OIr. és is used mostly in prepositional phrases like ar éis ‘after’. The Gen. sg. is probably attested only in the toponym Traig Ési. A trace of the same word in Brittonic could be the Toponym

13

Ad Ansam from Itinerarium Antonini. The PCelt. form can be derived from *pant-teh2, which presupposes a nasal present comparable to Lat. pando (see Hamp 1981). The metathesis *pat-n- > *pant- (> Lat. pando) should also be assumed for Celtic. For the semantic connection with the root *peth2- ‘spread out, broaden’, cf. Lat. passum ‘step’. The same PIE root may be attested in PCelt. *fatamā ‘palm of the hand’. REF: Hamp 1981, Irslinger 2002: 342f. *farawsyo- ‘temple’ [Noun] GOID: OIr. ara gl. tempus, arae [yo m] COGN: Gr. pareiaí [p] ‘cheeks’ (Aeolic parau ai) SEE: *fare ‘in front of’, *awsos ‘ear’ ETYM: A compound of *fare- ‘in front of’ < PIE *prH(i)- and the word for ‘ear’, PIE *h2ews- (se *awsos ‘ear’). Probable cognates are found in Gaul. toponyms such as Arausiō (today’s Orange). REF: LEIA A-82, Delamarre 51. *fare ‘in front of’ [Prep] GOID: OIr. air, ar [Aspirating, +Acc, +Dat.] W: MW ar-, er- BRET: MBret. er-, ar GAUL: Are-morici [Ethnonym] CELTIB: are-korata [Toponym](A 52) PIE: *prH(i) ‘in front of’ (IEW: 810-812) COGN: Skt. purā, Gr. pára, OHG furi ETYM: LEIA A-37 relates the Celt. forms to Gr. perì, Go. faur-, Lat. per- (PIE *peri- ‘near’), but this is less probable in light of the vocalism and the meaning; the development *rHV > PCelt. *arV is expected. The Insular Celtic forms are compatible with PCelt. *fari, which could be the original form of this preposition (Loc. sg. of an old root-noun?). However, Gaul. and Celtib. point to *fare, with the final *-e perhaps by analogy with *ande- < *ndhe. REF: LEIA A-37, GOI 497f., Delamarre 52, Villar 1997: 902, MLH V.1: 36f. *ferissā ‘religion, belief’ [Noun] GOID: OIr. iress [ā f] ‘belief’ PIE: *peri- ‘about, before’, *dheh1- ‘do, make, put’ ETYM: PCelt. *feri-ssā presupposes PIE *peri-dh1-teh2. EIEC derives OIr. iress from *peri-steh2- (from the root *steh2- ‘stand’), and compares Parthian parast ‘ardor’, but I believe the semantic connection with the root *dheh1- is better (cf. Skt. dhā man- ‘law’). The exact relationship of MW eirioes (GPC eiroes) [f] ‘faith, trust’ to OIr. iress is unclear to me. REF: EIEC 61, GPC I: 1197. *ferko- ‘perch’ [Noun], ‘speckled’ [Adj] GOID: OIr. erc [o m] ‘perch, salmon’, Ogam. ERCA W: MW erch ‘spotted’ PIE: *perḱ- ‘color’, *perḱ-no- ‘speckled (fish)’ (IEW: 821) COGN: Gr. perknós ‘freckled, red’, OHG ferhana ‘trout’. ETYM: The original meaning of this substantivized adjective, ‘speckled’, may be preserved in Ogam PN ERCA. REF: LEIA O-28, GPC I: 1229, EIEC 537, 634, EIEC 113, Ziegler 1994: 104. *ferkunyo- [Toponym] SEE: *kwerxt- ‘bush’ *fētu- ‘(grass-)land, territory’ [Noun] GOID: OIr. íath [u n] ‘land, territory’ PIE: *peyH-tu- ‘prairy, rich grassland’ (IEW: 793) COGN: Gr. poíē ‘grass’, Lith. píeva ‘meadow’ SEE: *fīweryon- ‘earth, soil’ ETYM: The name of the Gaulish tribe Heluētiī could be a compound *felu-ētyo- ‘Die Landreichen’ (see Delamarre 168). The PCelt. form is from PIE *peyh2-tu-. The laryngeal is reconstructed on the basis of the acute in Lithuanian, which means that this root has to be separated from *peyt- ‘feed’,

14

pace Malzahn 2011 (cf. *fitu- ‘food’). It is probably the same root as in PCelt. *fīweryon- ‘earth, soil’ < *piHwer-. I am not sure whether OIr. íth [u n] belongs here as well, but it could, in principle, be from *fītu- < *pih2tu-). REF: Delamarre 168, Widmer 2004: 19, McCone 1994: 114f., Irslinger 2002: 109, 165f., Malzahn 2011. *flikkā ‘(flat) stone’ [Noun] GOID: OIr. lecc [ā f] ‘(sepulcral) plate, stone, flat slab of rock’ W: MW llech [f] ‘slate, slab of stone’ BRET: MoBret. lec’h CO: Co. lehan GAUL: Are-lica [Toponym] PIE: *plkeh2 ‘flat surface’ (IEW: 831f.) COGN: Gr. pláks ‘flat stone’, ON flá ETYM: The source of the geminate in Celtic is unclear. Perhaps PCelt. *flikkā contains the velar suffix *-keh2 added to the root *plk-. REF: GPC II: 2123, Delamarre 201, de Bernardo Stempel 1999: 178, 508ff. *fot-ī- ‘throw’ [Vb] SEE: *fet-o- ‘fly’ *fotlo- ‘drink, act of drinking’ [Noun] GOID: OIr. ól, óol [o n] PIE: *peh3- ‘drink’ (IEW: 840) COGN: Lat. pōculum ‘chalice’, Gr. pósis ‘drink’ SEE: *fib-o- ‘drink’ ETYM: The vowel *o in Celtic is unexpected, as the PIE laryngeal should have yielded *a between consonants. It is probably due to an early analogy with the full grade (*eh3 > *ō > PCelt. *ā), or to vowel assimilation (*fatlo- > *fotlo-), or to the so-called ‘Wetter-rule’, by which laryngeals were lost before consonant clusters (*peh3-tl- > *potlo- > *fotlo-; note that this rule is not generally accepted). Original *peh3-tlo-would presumably have given OIr. **ál. REF: LEIA O-19, LIV 462f., de Bernardo Stempel 1999: 302, Zair 2010: 202. *frato- ‘grace, virtue, good fortune’ [Noun] GOID: OIr. rath [o n] W: OW rat [m], MW rat (GPC rhad) BRET: OBret. Rad-(uueten) [PN] CO: Co. ras SEE: *far-na- ‘bestow’ ETYM: These nouns are probably related to the root of *far-na- ‘bestow’ (OIr. ernaid, etc.). Gaul. PN Su-ratus may contain the same root. The PIE form (maybe an old passive participle meaning ‘bestowed, given’) would be *prh3tó-. This would first have yielded PCelt. *frātó-, and then *frato- by Dybo’s law. It is unclear whether OIr. ráth [ā f] ‘suretyship, pledge’ is connected with this root, but if it is, it might be from PCelt. collective *frātā ‘the sum of things given as a pledge’. REF: LEIA R-8, GPC III: 2995, Delamarre 255, de Bernardo Stempel 1999: 442, McCone 1996: 52, Irslinger 2002: 354, Stüber 1998: 156. *frēmo- ‘chief, prince’ [Noun] W: MW rwyf [m] (GPC rhwyf) BRET: MBret. roue [m] ‘king’ CO: OCo. ruy gl. rex, ruif ‘king’ PIE: *prey- ‘before, at’ (IEW 812) COGN: Lat. prīmus ‘first’, Gr. prín ‘before’, OCS pri ‘at’ ETYM: Campanile (1974: 91) derives the Brittonic forms from Lat. rēgem, which is unlikely: the accusative ending –em is never preserved in Brittonic as –m > -f. Haarmann (1970) does not list rhwyf among Latin loanwords in Welsh. The Celtic forms are derived from < *prey-mo-, cf. Lat. prīmus <

15

*prey-smo-. Possible cognates in Gaulish include the Ethnonym Remi (> Reims) and NPs Remus, Remos, etc. REF: GPC III: 3115, Campanile 1974: 91, Delamarre 257. *fritu- ‘ford’ [Noun] GOID: MIr. Humar-rith [Toponym] W: OW rit, W rhyd [m] BRET: OBret. rit gl. uadum CO: OCo. rid gl. vadum, Co. red GAUL: Ritu-magus [Toponym] PIE: *prtu- ‘ford’ (IEW: 817) COGN: Lat. portus ‘port’, OE ford SEE: *foro- ‘edge, limit’ ETYM: The fact that MIr. Humar-rith (with unclear first element) denotes a ford is clear from the context (Humarrith ainm ind átha sin ‘the name of the ford was Humarrith’, LU 5812). REF: LEIA R-34, GPC III: 3126, Delamarre 259, Irslinger 2002: 123, Sims-Williams 2006: 103f. *friyo- ‘free’ [Adj] W: MW ryd, rydd (GPC rhydd) BRET: OBret. rid CO: OCo. benen rid gl. femina PIE: *priHo- ‘dear, free’ (IEW: 844) COGN: Skt. priyá-, Go. freis ‘free’, OCS prijati ‘be appealing to’ ETYM: Some Gaul. names with the element Rio- (e.g. Riotalus) might be related to this root, rather than to *rīg- ‘king’ (Delamarre 258). REF: GPC III: 3127, EIEC 214, 358, Delamarre 258f., Benveniste 1969: 325ff., Zair 2010: 228. *fowl-wo- ‘ashes, dust’ [Noun] W: MW ulw [p m] PIE: *pewl- ‘dust (?)’ COGN: Lat. puluis ‘dust’ ETYM: The etymology proposed here is possible only under the assumption that Lat. puluis is not etymologically related to palea ‘chaff’ < PIE *pelh1- (as assumed by de Vaan 2008: 498). Note that puluis and MW ulw are closer semantically and that both forms can go back to an u-stem with the Nom. sg. *powlu- (from which we have MW ulw, with usual change of adjectival u-stems to wo-stems in Brittonic), Gen. sg. *pulw-os, from which Lat. puluis is ultimately derived (with the transfer to neuter s-stems, perhaps by analogy with cinis, cineris ‘ashes’. A trace of an etymologically related Gaulish word may be preserved in French dialects, e.g. luvre (Poitoux) ‘ashes’, orve (Champagne) ‘flour’, etc. REF: GPC IV: 3701, Delamarre 324, Hamp 1986b. *futro- ‘ill’ [Adj] GOID: MIr. othar [o] ‘ill’, othar [o n, later m] ‘illness, pain’ PIE: *pewH- ‘rot, decay’ (IEW: 849) COGN: Lat. puter ‘rotten’, Skt. puvas- ‘pus’, Gr. pýos ‘pus’, OIc. fúinn ‘rotten’ ETYM: Short *u in Celtic is presumably due to Dybo’s shortening in pretonic position, i. e. PIE *puHtró- > PCelt. *futro-. Otherwise, the laryngeal may have been lost because of the so-called ‘Wetter-rule’, which is not generally accepted (cf. *fotlo- ‘drink’ which may be from *peh3-tlo-). REF: LEIA O-36, Zair 2010: 202f. *gweltā ‘grass’ [Noun] GOID: OIr. geilt ‘grazing’ [ā, f] W: MW gwellt [m] ‘grass’ BRET: OBret. gueltiocion gl. ‘fenosa’, MBret. gueautenn [Singulative], MoBret. geot [Collective] CO: MCo. gwels ‘grass’ SEE: *gwel-o- ‘graze’ REF: GPC II: 1632f., KPV 372, DGVB 188, LP 28, Sims-Williams 1982: 228f., Zair 2010: 244f.

16

*gan-yo- ‘be born’ [Vb] GOID: OIr. gainidir, gainithir, -gainedar; -genadar [Subj.]; -gignethar [Fut.]; génair [Pret.] W: MW geni; ganaf [1 s Pres.] BRET: MBret. guenell CO: Co. genys [Part. Pass.] PIE: *enh1- ‘beget’ (IEW: 373ff.) COGN: Skt. jani-, Lat. gigno, nāscor, OE cennan SEE: *geno- ‘family, gens’ ETYM: OIr. -gainethar, -gainedar has the regular reflex of syllabic *-n- before any consonant except a laryngeal. Two developments are possible, either *nh1yetor > *gnyetor (with an early loss of the laryngeal before *y by Pinault’s law) > *ganyetor > *ganitor > gainethar, or *nh1yetor > *gnHitor > *ganHitor > gainethar. OIr. gein [n n] ‘birth, conception’ is another derivative from the same root. It presupposes PCelt. *genan < *genh1en (an archaic-looking neuter n-stem). The root is also attested in the e-grade in Celtiberian as the second element of the compound PN Mezu-kenos (= OIr. Midgen), and in the zero-grade in Ogam Irish PN DALA-GNI [Gen s]. REF: GPC II: 1380, KPV 327ff., LIV 163ff., McCone 1994: 70, MLH V.1: 251. *gat-o- ‘let, allow’ [Vb] W: MW gadu ‘permit, let, allow’ CO: Co. gasa, gase PIE: *heh1- ‘leave, let’ (IEW 418f.) COGN: jáhāti ‘leaves’, Gr. khē ros ‘bereaved, widowed’ ETYM: A very tentative etymology. The PCelt. present stem of this verb cannot be established on the basis of Brittonic evidence alone, but a thematic verb is likely. The Celtic form looks like a deadjectival verb built from the to-participle *hh1-to-, but the participle itself is unattested in Celtic. REF: GPC II: 1367, LIV 173. *genetā ‘girl’ [Noun] W: W geneth [f] GAUL: geneta, genata, gnata PIE: *enh1- ‘bear, engender, generate’ (IEW: 373ff.) COGN: Lat. -genitus in primo-genitus ‘first born’ SEE: *genos- ‘family’ ETYM: W geneth points to a geminate *genettā; this PCelt. noun is derived with the suffix *-eto- (cf. Osc. genetaí ‘daughter’ [Dat. sg.], cf. also OIr. geined, geinit [o m] ‘creature’ < *geneto- and OIr. aicned [o n] ‘nature, essence, spirit’ < *ad-geneto-. OIr. ingen [ā f] ‘daughter’ < *eni-genā (Ogam INIGENA) is another formation from the same root, cf. also Gaul. (Larzac) andognam ‘born inside (the family)’ and PN Andegenus. OIr. PN Sogen (Ogam Gen. SOGINI) is presumably from *su-enh1o- ‘well-born’ (Skt. sujana-). REF: GPC II: 1393, LIV 163f., Delamarre 177, Uhlich 2002: 422, Wodtko et alii 2008: 142f. *geno- ‘family, gens’ [Noun] GOID: OIr. Éo-gan [PN], Ogam INI-GENA ‘daughter’ W: OW Mor-gen [PN] BRET: OBret. gen gl. ethnicus, MoBret. genel ‘generate’ GAUL: Ad-genus, Cintu-genus [PN] PIE: *enh1-os ‘family, clan, descendants’ (IEW: 373ff.) COGN: Skt. jánas-, Gr. génos, Lat. genus SEE: *gan-yo- ‘be born’ REF: DGVB 174, Delamarre 177f., Deshayes 2003: 270, Ellis Evans 1967: 204, Ziegler 1994: 103. *gessī ‘taboo, prohibition’ [Noun] GOID: MIr. geis [?ī f] PIE: *ghed- ‘seize, take’ (IEW: 437f.) COGN: Lat. prae-hendo, Gr. khandánō SEE: *gan-d-o- ‘take place’

17

ETYM: The MIr. word can be derived regularly from *ghed-ti-h2, but the semantic development (’taking’ > ‘prohibition, taboo’) is somewhat difficult (see Hamp 1981). Moreover, it is by no means certain that the original formation was an ī-stem. If an ā-stem is original in Irish (as supposed by Irslinger 2002: 348), we might reconstruct PCelt. *gissā (> OIr. *gess by lowering), and derive this word from PIE *ghidh-teh2 ‘what is desired’ (cf. PIE *gheydh- > Lith. geidžiù ‘desire’, OCS žьdati ‘wait’, IEW 426). The same root is probably attested in PCelt. *gēstlo- ‘surety, pledge, hostage’. REF: Hamp 1981, Irslinger 2002: 348ff. *glamā ‘cry, shout’ [Noun] GOID: MIr. glám [ā f] ‘outcry, clamour, satire’ PIE: *gla-h- ‘cry, shout’ (IEW: 350f.) COGN: Skt. garh-, OHG klaga, Germ. klagen ETYM: The comparison with OHG klaga led IEW to derive MIr. glám from *glaxsmā, but the lenited mh in Modern Irish glámh ‘satire’ shows that the correct PCelt. reconstruction is *glamā. Other possible comparanda include OIc. kall ‘cry’ and Russ. gólos ‘voice’, which could be derived from *golHso- (with the laryngeal lost by de Saussure’s rule), but there is no independent evidence for a laryngeal in that root (see PCelt. *galwo- ‘call’. PCelt. *glamā should probably be derived from an onomatopoetic root, so these correspondences could be accidental. REF: Mayrhofer I: 475f., Zair 2010: 106. *glasto- ‘green, blue’ [Adj] GOID: OIr. glas [o], Ogam GLASI-CONAS [PN] W: OW glas, MW glas BRET: MBret. glas CO: OCo. glesin gl. sandix GAUL: glastum ‘Isatis tinctoria’ (Pliny) ETYM: For the apparent preservation of *-st- in Gaul., cf. Schrijver 1995: 402. In principle, these forms could represent PIE *hlh3-stó- > *glāsto- > *glasto- (by Dybo’s shortening). The PIE root would have been *helh3- ‘yellow, green’ (IEW 429ff.), as in *gel(w)o-. Cf. also MHG glasta ‘shine’, which may reflect a prehistoric borrowing from Celtic (if it is related at all). The form of the (?composite) suffix *-sto- is unclear, but I do not believe we are dealing with an old compound *hlh3-sth2ó- ‘standing in green(?)’ as suggested by Hill (2003: 274). REF: GPC II: 1401, Delamarre 180, Falileyev 61, Campanile 1974: 49, Deshayes 2003: 273, Delamarre 180, McCone 1996: 99, De Bernardo Stempel 1999: 271, Schrijver 1995: 402, Irslinger 2002: 427, Hill 2003: 274, Zair 2010: 104f. *gliwo- ‘valiant, brave’ [Adj] GOID: MIr. gleo W: MW glew ‘valiant, bold, fierce’ CO: Co. glew ‘fierce’ ETYM: The original stem formation of OIr. gleo (also spelled gleó) is diffixult to ascertain. It is inflected as a feminine dental stem (Gen sg. glíad), but it is also attested as a feminine ā-stem (Gen. sg. glee, glé), and it is assumed here that this formation is original. It represents the substantivized adjective *gliwā ‘valiance, bravery’ > ‘fight, combat’. REF: GPC II: 1497, de Bernardo Stempel 1999: 220. *gloydo- ‘glue, lime’ [Noun] GOID: MIr. glóed (DIL gláed) W: MW glud [m] BRET: MBret. glut, MoBret. glud CO: OCo. glut gl. gluten, MoCo. glüs SEE: *gli-na- ‘glue’ ETYM: The gender and stem-formation of MIr. gláed are uncertain. The Brit. forms point to word-final *-t rather than *-d. It is possible that they were actually borrowed from Lat. gluten (Haarmann 1970: 122), or that their Auslaut was influenced by that word. Cf. also Fr. glaise ‘clay’ which may be from Gaul. *glēssā < *gleyd-tā (Gamillscheg 480). REF: GPC II: 1412, Deshayes 2003: 276, Zair 2010: 314.

18

*gloysto ‘brightness’ [Noun] GOID: OIr. glésse [yā f] ‘brightness’ W: W glwys ‘beautiful, pleasant, pure’ SEE: *gloywo-, *glēwi- ‘liquid, clear’ ETYM: It is assumed here that the adjective found in W is primary (PCelt. *gloysto-), while the OIr. noun is derived from a secondary abstract noun (*gloystyā). REF: GPC II: 1413. *gnāto- ‘known, usual’ [Adj] GOID: OIr. gnáth [o] W: MW gnawt (GPC gnawd) BRET: OBret. gnot ‘usual’ PIE: *enh3- ‘know’ (IEW: 373ff.) COGN: Gr. gignō skō, Lat. co-gnōsco, OCS znati, Lith. žinóti ETYM: Certain Gaulish compounds with -gnatos can be understood both as ‘known’ < *neh3tos and as ‘born’ < *nh1tos (Delamarre 181f.). Note that the pre-form *neh3tos is more probable than *nh3tós for PCelt. *gnāto-, since we would expect the operation of Dybo’s law if the latter were true. Next to adjectives in *-to- we also find nouns in *-sto- from this root (a productive pattern in Celtic, see Greene 1965): OIr. gnás [ā f] ‘custom, usage, intercourse’, W gnaws (GPC naws) < PCelt. *gnāsto- < *nh3sto- (cf. Go. kund vs. kunst). REF: GPC II: 1415, EIEC 337, Delamarre 181f., de Bernardo Stempel 1999: 337, 440, Greene 1965, Irslinger 2002: 267f., 415f., Zair 2010: 96. *gnāw(y)o- ‘clear, manifest’ [Adj] GOID: MIr. gnóe, gnó [yo] ‘beautiful, fine, exquisite’; [yā f] ‘beauty, distinction, knowledge’ W: MW gno ‘evident, clear, manifest’ BRET: MBret. gnou ‘manifest, evident’ SEE: *gnāto- ‘known, usual’ ETYM: MW gognaw ‘provoking, exciting, fierce’ is probably from the same root (with the prefix *ufo-). The semantic develpoment was probably from ‘known’ to ‘obvious, evident’ and ‘clear’, and then in Goidelic to ‘beautiful, fine’. REF: GPC II: 1415, de Bernardo Stempel 1999: 220, Zair 2010: 121f. *gnūsti- ‘face, chin’ [Noun] GOID: OIr. gnúis [i f] ‘face, countenance’ W: MW gnis [m] ‘jaw’, chin, face’ ETYM: This noun is probably related to PCelt. *genu- ‘jaw’, but it is unclear exactly how. It may be based on the original neuter plural *g(e)nū < *henuh2, but the origin of the suffix *-sti- is unclear. REF: GPC , Irslinger 2002: 411, 418, Zair 2010: 162. *gowstā ‘chance, possibility’ [Noun] GOID: OIr. gúas ‘danger, peril, fear, chance, possibility’ PIE: *ews- ‘taste, choose’ (IEW: 399f.) COGN: Skt. jus- ‘like, enjoy’, Gr. geúomai ‘taste’, Go. kiusan ‘check’ SEE: *gus-o- ‘choose’, *gustu- ‘excellence, force’ ETYM: Although the earliest meaning of OIr. gúas is ‘danger’, I believe this is not the basic meaning. Rather, one has to start from ‘chance, possibility’ (the meaning attested in MIr. according to DIL), which developed, in turn, from ‘choice’, which is obviously related to the meaning of the verb from the same root, PCelt. *gus-o- ‘choose’. A similar semantic development is attested in Croatian, cf. Croat. iskušenje ‘temptation, peril’ from the verb kušati ‘check, taste’. A derivative from the same root is OIr. gúasacht [ā f] ‘danger’, perhaps attested in Ogam as Gen. sg. GOSSUCTTIAS. REF: Ziegler 1994: 185, de Bernardo Stempel 1999: 444, Irslinger 2002: 105, 108. *gressā ‘group of guests’ [Noun] GOID: MIr. gress [ā f] W: W gresaw ‘welcome, greeting of hospitality’ (GPC greso, gresaw)

19

ETYM: The direct reflex of PCelt. *gressā is reflected in MIr. gress, while W gresaw represents a derivative. The PIE etymology is very uncertain. The connection with OE grétan ‘greet’ Germ. Gruss ‘greeting’ is formally impossible, and the derivation from the root *gres- ‘graze, devour’ (Skt. grásate, LIV 170, IEW 404) is semantically difficult, and this root is otherwise unatested in Celtic. A derivation from the PIE root *ghreydh- ‘go, walk’ (see PCelt. *gri-n-d-o- ‘follow, drive’) would imply that the correct pre-form of MIr. gress is *grissā (< *ghrdh-teh2). Again, this would involve complex semantic changes (‘a group of travellers’ > ‘a group of guests’). A similar semantic development (a noun meaning ‘guest’ from the root meaning ‘to go’) can be observed in PCelt. *oyget- ‘guest’. REF: GPC II: 1529, Irslinger 2002: 350. *gritu- ‘cry, roar, vibration’ [Noun] GOID: OIr. grith [u m] ‘vibration, panic, cry’ W: MW gryt [m] (GPC gryd) ‘shout, yell, call, cry] ETYM: The derivation from the root of *gar-yo- ‘call’ is improbable, since that root is here reconstructed with a laryngeal. The comparison with onomatopoetic forms such as Lat. gingrīre ‘cry (of geese)’ found in IEW 384 is not illuminating, and the connection with the Germanic words such as OHG scrīan ‘cry’ would be possible only under the difficult assumption that the Germanic forms go back to *sg- > *sk- (with s-mobile). Therefore, the PIE etymology of these Celtic forms must remain uncertain. REF: GPC II: 1539, Irslinger 2002: 104. *gulbV-, *gulbīno- ‘beak’ [Noun] GOID: MIr. gulba [n f] ‘beak, jaw’, gulban [o m] ‘beak, sting’ W: OW gilbin [m and f] ‘point’, MW gylfin ‘beak’ BRET: OBret. golbin, golbinoc gl. ac rostratam; MBret. golff, MoBret. golv ‘without tail’ CO: OCo. geluin gl. rostrum GAUL: *gulbia > Lat. gulbia (Isidore of Seville) ETYM: Cf. also Brit. Re-gulbium (name of a promontory). A persuasive IE etymology has not been proposed. A connection with Gr. gláphō ‘scrape up, dig’ (< PIE *glbh-, IEW 367) is improbable, since if does not explain PCelt. *u. The similarity with the Balto-Slavic words for ‘swan’ (Lith. gul bė, OPr. gulbis, Russ. kolpíca ‘young female swan’, Smoczyński 2007: 209) is probably accidental. This PCelt. etymon was probably borrowed from some non-IE language. REF: GPC II: 1794, DGVB 178, Delamarre 184, Campanile 1974: 48, Falileyev 61, Deshayes 2003: 278, Stüber 1998: 110, de Bernardo Stempel 1999: 118, 459. *gustu- ‘excellence, force’ [Noun] GOID: OIr. guss [u m], Ogam CUNA-GUSSOS W: OW Gur-gust [PN], W gwst [f and m]; gystion [p] ‘pain, endurance, patience, difficulty’ GAUL: gussou [Dat. s] ‘? force’ (Lezoux) PIE: *us-tu- ‘choose, taste’ (IEW: 399f.) COGN: Lat. gustus ‘tasting’, OE cost ‘choice, excellence’ SEE: *gus-o- ‘choose’, *gowstā ‘chance, possibility’ ETYM: For the apparent preservation of *st in British, see Schrijver 1995: 412. The semantic developent was from ‘choice’ to ‘excellence’ and ‘force’, cf. *gus-o- ‘choose’. It is unclear whether MIr. gus ‘danger’ is the same word originally (it is a hapax in the glosses). I do not find persuasive its derivation from the root *hewd- ‘pour’ (Irslinger 2002: 107f. The semantic connection would be in the concept of ‘pouring of blood’). REF: GPC II: 1742f., EIEC 566, Delamarre 184, Lambert 1994: 146f., de Bernardo Stempel 1999: 290, Hamp 1986, Schrijver 1995: 412f., Ziegler 1994: 107, Irslinger 2002: 105ff. *i-tro- ‘ferry, boat’ [Noun] SEE: *i- ‘go’ *kwākwo- ‘everyone’ [Pron] GOID: OIr. cách W: OW paup, MW pawb BRET: OBret. pop, MBret. pep

20

CO: Co. pup, pop, pep GAUL: papi [Gen s], papon [Acc s] (Lezoux) PIE: *kwo-h3kwo- (IEW: 645) COGN: OCS kakъ ‘qualis’ SEE: *kwēs ‘who’ ETYM: Unstressed forms of this word serve as adjectival pronouns meaning ‘every’ (OIr. cach, cech, MW pob). The PIE form *kwo-h3kwo- is a compound consisting of the interrogative pronoun stem (*kwo-) and the stem of the word for ‘eye’ (*h3ekw- > Lat. oculus etc.). The original meaning could have been ‘whatever the eye (sees)’. Another possibility would be to reconstruct a reduplicated *kweh2-kwo- (cf. long ā in Lat. quālis). As Michiel de Vaan points out to me (p.c.) reduplicated interrogatives often yield indefinite pronouns (cf. Lat. quisquis). REF: LEIA C-3, GPC III: 2703, EIEC 457, Delamarre 246f., Falileyev 128. *kwenkwekont- ‘fifty’ [Num] GOID: OIr. cóica [m nt] W: OW pimmunt, MW pymmwnt SEE: *kwenkwe ‘five’ BRET: OBret. pimmont ETYM: The OIr. form presupposes a haplology (in allegro-speech?) *kwenkwekont- > *kwenkont-. A similar process would account for the Brittonic forms. The second element of this compound, *-kont-, is presumably from *dkomt- (from the PIE numeral *deḱm ‘ten’, cf. the same element in Gr. pentē-konta ‘fifty’). REF: LEIA C-143f., DGVB 286. *kallī- ‘wood, grove’ [Noun] GOID: OIr. caill [?ī f] W: MW celli [f] ‘grove, copse’ BRET: MoBret. Quille-vere [PN] CO: OCo. kelli gl. nemus ETYM: The geminate *-ll- is certainly the result of assimilation, but it cannot be ascertained which consonant was originally involved as the second member of the cluster. The obvious solution, PIE *kldo- (Gr. kládos ‘branch’, ON holt ‘woody hill’), will not do, since we would expect PCelt. *klido- (cf., e.g., PIE *wld- > PCelt. *wlidā ‘feast’). Moreover, the cluster *ld would have been preserved in Celtic. A derivation from the root *kelH- ‘cut’ (IEW 545ff.) remains a possibility, but details are obscure. REF: LEIA C-13, GPC I: 459. *kani- ‘good, nice’ [Adj] GOID: OIr. cain [i] W: MW cein (GPC cain) BRET: MBret. quen, MoBret. ken ETYM: The Brittonic reflexes point to the yo-stem (*knyo-), but the OIr. i-stem adjective is probably more archaic. These Celtic words are often derived from PIE *ken- ‘begin’ (IEW 563f., cf. Gr. kainós ‘new’, OCS na-čęti ‘begin’), but the semantic difference is too great. Since a convincing IE etymology is lacking, PCelt. *kani- may have been borrowed from some non-IE language. REF: GPC I: 390, LEIA C-15, Deshayes 2003: 383, LIV 351. *kart-ā- ‘cleanse’ [Vb] GOID: OIr. cartaid, -carta ‘sends, dispatches, expels, drives off’ W: W carthu ‘cleanse, purge’ BRET: MoBret. karza ‘cleanse’ SEE: *skara- ‘divide, separate’ ETYM: The semantic reconstruction is the key to the etymology of this word. If the meaning in Brittonic is original, the semantic evolution in Goidelic was from ‘cleanse’ to ‘expel’ and ‘drive off’. In that case, the PIE root may be the same as in *skara- ‘divide, separate’ (which has s-mobile). REF: LEIA C-43, GPC I: 433. *kato- ‘wise,, holy, pure’ [Adj]

21

GOID: OIr. cath [o] (DIL cáid [i]), cád, cad PIE: *ḱeh3- ‘sharpen’ (IEW: 542) COGN: Skt. śitá- ‘sharpened’, Lat. catus ‘clever’ ETYM: The OIr. word is spelled differently as cath, cad, cáid and cád (see DIL for attestations). It is unclear whether the variability of the length of the root vowel could represent old ablaut, or are we dealing with two originally different words. As in Latin, the Celtic word presupposes the semantic development of ‘sharp’ to ‘wise, clever’. A possible cognate is also Gallo-Lat. caddos ‘holy’ (CGL V 493, 30). The attribution to the PIE root *ḱeh3- ‘sharp’ is possible; however, the identity of the laryngeal is surmised chiefly on the basis of Lat. cōs, cōtis ‘whetstone’, and the etymology of this word is not completely certain. REF: LEIA C-10, Delamarre 96, Schrijver 1991: 91, de Vaan 2008: 99. *kēro- ‘dark brown’ [Adj] GOID: OIr. cíar [o] PIE: *ḱey-ro- ‘horn’ (IEW: 540f.) COGN: OIc. hárr ‘grey-haired’, OE hār ‘grey, grey-haired’, Russ. sěryj ‘grey’, Cz. šerý SEE: *kiwo- ‘fog’ ETYM: MIr. cíarann ‘beetle’ is a derivative from the same root. Because of initial š- in Czech, we must assume that the Slavic forms go back to *x-, which was palatalized before *ě. Because of this, the root is sometimes reconstructed as *ḱh1ey-, under teh assumption that *ḱH yields Slavic *x-. This would make easier the comparison with PCelt. *kiwo- ‘fog’, which is from PIE *ḱeyH-. However, *x- in Slavic can also be from *sḱ-, i.e. the Slavic form can reflect the root with s-mobile. REF: LEIA C-95, GOI 204, Derksen 2008: 447, Zair 2010: 130, Lubotsky 1989: 56, Zair 2010: 130. *kitu- ‘shower (of rain or snow)’ [Noun] GOID: MIr. cith [u m] PIE: *ḱey- ‘lie’ (IEW: 542) COGN: Skt. śáye ‘lies’, Gr. kei mai, OPr. keytaro ‘hail’ SEE: *ki-yo- ‘fall, cry’ ETYM: This is a very tentative etymology, as it relies on a single form in MIr. and presupposes a semantic development from ‘lie’ to ‘fall’ (and then to ‘shower’ in Goidelic). LIV posits a different PIE root, *ḱey- ‘fall’ and compares Ved. áva śīyate ‘falls’, but the similarity with *ḱey- ‘lie’ is probably not accidental. Probable cognates in Brittonic include W cwyddo ‘fall’, MoBret. koezaff ‘fall’ < *kē-de- < ? PIE *ḱey-dhe-. REF: LEIA C-108, LIV 321, Schrijver 1995: 224, Irslinger 2002: 91. *klamo- ‘sick, suffering from leprosy’ [Adj] GOID: OIr. clam [o] ‘leprous’ W: MW claf ‘sick, ill, leprous’ BRET: MBret. claff, MoBret. klañv CO: OCo. claf gl. eger ETYM: The a-vocalism of this etymon points to non-IE origin. In Brittonic, there is also a derivative *klamito- ‘sickness’ > OCo. clevet gl. morbus, MoBret. klen ved [m] ‘sickness’, W clefyd. REF: LEIA C-112, GPC I: 487, LP 54, de Bernardo Stempel 1999: 45, Campanile 1974: 27, Deshayes 2003: 394. *klisso-, *klissu- ‘feat’ [Noun] GOID: OIr. cles [o and u m] ‘feat, trick, strategem’ PIE: *kleys-d- COGN: Skt. krīḍati ‘plays, dances’ ETYM: If OIr. cles is related to Skt. krīḍati, the PIE root should be reconstructed as *kleys-d-. PCelt. *klisso- < *klisd-to- preserves the zero-grade of the root from the past participle. However, Skt. krīḍati has also been related to ON hrista ‘shake’, which must be from PIE *kreys-d-. Another possibility would be to relate OIr. cless to Skt. śreṣ- ‘hang’ (Mayrhofer II: 670f., LIV 333), in which case we should reconstruct PCelt. *klisso- (< PIE *ḱlis-so-), or *klisto- (< PIE *ḱlis-to-). Another possibility is that cless is related to MIr. cluiche ‘play, game’ (perhaps from the same onomatopoetic

22

root as OE hlihhan ‘laugh’, OHG hlahhan), in which case it would be derivable from *klixso- < *klk-so-. REF: LEIA C-117, Irslinger 2002: 147f. *klito- ‘warm’ [Adj] W: MW clyd ‘warm, sheltered’ PIE: *ḱlto- ‘warm’ (IEW: 551) COGN: Lith. šil tas, Lat. calidus, caleo ‘be warm’ ETYM: Formally, PCelt. *klito- is a past participle of the verbal stem *ḱel-, which is attested in Lat. caleo. I find it improbable that OIr. clith [i] ‘thick, pressed, narrow (of cloth)’ is related to this root because the meanings are very different. REF: GPC I: 515, Irslinger 2002: 200. *klitro- ‘post’ [Noun] GOID: OIr. clithar [o m] ‘shelter, cover, protection’ W: MW cledr [f] ‘rod, stave, pole, rail, palm of hand, help’ BRET: MBret. klezr, klezren, MoBret. klerenn ‘wooden post’ ETYM: Derived from the same root as *klit- ‘pillar, post’, with the suffix *-ro-. The development of meaning in Goidelic may have been influenced by *klito- ‘warm’ (MW clyd). SEE: *klit- ‘pillar, post’ REF: LEIA C-120, GPC I: 493, Hamp 1982a, Irslinger 2002: 51. *kliyaro- ‘lukewarm, tepid’ [Adj] W: MW claear, claer ‘warm, gentle’ BRET: MoBret. klouar ‘tepid, kind, gentle’ CO: Co. clor, clour ‘gentle’ PIE: *kleyH- ‘warm’ COGN: OHG lao, Germ. lau, OE hlēowe ETYM: The Germanic forms represent the zero-grade of the PIE root (*kloyH-wo-), while the Celtic forms may be derived from the zero-grade (*kliH-), with the Celtic suffix *-aro-. The connection with Gr. khliarós ‘lukewarm’ is formally impossible. REF: GPC I: 486, LP 17, Deshayes 2003: 400. *knidā ‘wound’ [Noun] GOID: OIr. cned [ā f] COGN: OIc. hníta ‘strike’, Gr. knízō ‘scratch, rub’ PIE: *kneyd- ‘scratch’ (IEW 561f.) ETYM: MIr. cnes [o m] ‘skin, surface, flesh’ can represent the to-participle of the same PIE verb (*knid-to- > PCelt. *knisso-, cf. also Co. cnes ‘skin’), but the verbal forms are not attested in Celtic. REF: LEIA C-131, Irslinger 2002: 263f., LIV 366. *knisso- ‘skin, surface’ [Noun] SEE: *knidā ‘wound’ *kolanī- ‘body’ [Noun] GOID: OIr. colainn [i f] ‘body, flesh, corpse’ W: MW kelein [f] (GPC celain) ETYM: The Welsh forms point to PCelt. *kolanī , while the Irish word is from an i-stem (*kolani-). The IE etymology suggested by IEW (924) is dubious: the Celtic forms are derived from the root *(s)kel- ‘cut’ (see *skoltā ‘fissure, cleft’); the semantic development would be from ‘flesh, cut meat’ to ‘dead body, corpse’, cf. OIc. hold ‘flesh’ vs. OE hold ‘corpse’ and OE holdian ‘cut up, but the word-formation of the Celtic words is unclear even if that etymology is correct. REF: LEIA C-156 , GPC I: 455, Schrijver 1995: 95, Zair 2010: 313. *kom-rūno- ‘secret, confident’ [Adj] GOID: MIr. comrún, cobrún ‘joint secret’ W: MW cyfrin

23

BRET: MBret. queffrin ‘mistery, secret’ ETYM: In Gaulish, one compares the NPs Cobrunus, Cobruna, etc. REF: LEIA C-, GPC I: 727, Delamarre 123. *koynnV- ‘foam’ [Noun] GOID: OIr. cóenna gl. putamen, cáenna ‘foam’ [k f] BRET: OBret. ceuni gl. muscus ETYM: A difficult etymology since the forms are poorly attested. The comparison with Lat. caenum ‘mud, filth, slime’ is formally impossible. Better is the connection with Swedish (dial.) hven ‘low, marshy field’, OIc. -hvein (in placenames), but here the meanings are quite different. REF: LEIA C-7. *krid- ‘shrink, grow thinner’ [Vb] GOID: MIr. credb(a) ‘shrinkage’, credba(ig)id, -credbaigi ‘contract, grow thinner’ W: MW cryddu ‘shrink’ BRET: MBret. crezz, MoBret. krez ‘stingy’ PIE: *(s)krdh- COGN: Skt. kṛdhu- ‘short, shortened, small’, Lith. skursti ‘lack’ ETYM: MIr. credb can be from *kridwV (stem and gender are unknown). MIr. cres [o] ‘small, narrow’ could also, in principle, be from the same root (PCelt. *krisso- < *krdh-to-). REF: LEIA C-227, GPC I: 621, Deshayes 2003:, de Bernardo Stempel 1999: 216. *kruw(y)o- ‘stable, enclosure’ [Noun] GOID: OIr. cró [o m]; cróe, cróa [io m]’enclosure, shed, sty’ W: MW creu [m] ‘shed’ (GPC crau, craw) BRET: OBret. crou gl. hara .i. stabulum porcorum, MBret. kraou CO: Co. crow ‘hut’ PIE: *krewh1- ‘hide’ COGN: OCS kryti ‘hide’, OE hrēodan ‘cover’ ETYM: The PIE etymology of these words is uncertain because of the semantics of the Celtic cognates, which points to the PCelt. meaning ‘round, circular, enclosure’ (rather than ‘covered, roof’ which seems to be primary in Slavic and Germanic). Greene 1983 relates the Celtic words for ‘enclosure’ to OIr. cruind, W crwnn ‘round’ (PCelt. *krundi-). However, this is difficult from the formal point of view (*-ndi- does not seem like a possible suffix). Likewise, OCS krěpъ ‘strong’, OIc. hraefa ‘endure’ and OE hróf ‘roof’ (PIE *kreh1po-/ *kroh1po-) cannot be related, since PCelt. *krāpo- would have given OIr. **crá, MW **kraw. Go. hrot ‘roof, house’, ON hrót, and OCS krada ‘pile of logs, pyre, altar’ are probably from same root with a dental suffix (*kreh1-do-), while we probably have *kreh1to- in OHG and OFris. rāza ‘honeycomb’. REF: LEIA C-240, GPC I: 582, DGVB 123, EIEC 217, Schrijver 1995: 343, de Bernardo Stempel 1999: 25, Greene 1983, Stalmaszczyk & Witczak 1995, Zair 2010: 227f. *ladano- ‘dumb’ [Noun] GOID: OIr. ladan [o] PIE: *leh1d- ‘tire, let’ COGN: Lat. lassus ‘tired’, Gr. lēdei n ‘be tired’, Alb. lodhem ‘be tired’ ETYM: This etymology is quite uncertain, since the meaning of OIr. ladan is quite different from that posited for PIE *leh1d-, but the semantic development ‘tired’ > ‘useless’ > ‘dumb’ is conceivable. If the etymology is correct, ladan can be from the zero-grade of the root *lh1d- and the Celtic suffix *-ano- (the PIE suffix *-no- generalized after roots in a laryngeal). Gaul. PN Ladanus can be plausibly connected with OIr. ladan, but as all etymologies of proper names this has to be taken cautiously. REF: Delamarre: 194, LIV 400, Zair 2010: 76. *lam-yo- ‘dare’ [Vb] GOID: OIr. ro-laimethar; ro-lamathar [Subj.]; ro-lilmathar [Fut.]; lámair [Pret.]; ro-lét [Pret. Pass.] W: MW llafasu, llyfasu, llyfasel BRET: MBret. lafuaez CO: Co. lauasos ‘to be allowed’ PIE: *h3lem- ‘to tire (oneself), to break (intransitive)’ (IEW: 674)

24

COGN: Gr. nōlemés ‘untiringly, unceasingly’, OHG lam ‘lame’, Lith. lémti ‘to ordain’, OCS lomiti ‘break’ ETYM: OIr. ro-laimethar < *fro-lam-yo- (the simplex verb is unattested in Goidelic). The acute in Lith. lémti (3sg. lémsta) must be due to metatony, which is expected in sta-presents. LIV (412) reconstructs the root without the initial *h3-, obviously believing that Gr. nōlemés is not related. In any case, the semantic variation of the reflexes of *h3lem- is considerable, so this etymology is not beyond reproach. This also applies to the meanings in Celtic. Perhaps the evolution was from ‘to tire’ to ‘to attempt’ and, finally, to ‘to dare’. MIr. laime ‘axe’ may be from the same root (*lam-yo- or *lam-yā?), but it is very poorly attested and its exact stem formation and gender are unknown. MIr. lem [o] ‘soft, impotent’ may, in principle, be from the same root in the e-grade (*lem-o-), but this is also very uncertain. REF: KPV 446f., GPC II: 2086, LIV 412, LP 376, EIEC 81, Stüber 1998: 135, Zair 2010: 115. *lēko- ‘wretched, pitiful’ [Adj] GOID: OIr. líach [o] ETYM: The etymology given by IEW 667 compares OIr. líach to Gr. loigós 'ruin, mischief, death, plague', Lith. ligà 'illness', Alb. lig 'bad' is difficult, since Celtic points to word final *-k, while the other languages have *-g. The comparison is possible only if one reconstructs a root PIEnoun *lōyg-s /loyks/, in which the velar was devoiced before final *-s; the resulting root-final *-k could have been generalized throughout the paradigm in Celtic, which subsequently formed a thematic adjective *leyko- > *lēko- from that root. Greek preserves the stem *loyg- from the PIE accusative (*loyg-m), while Lith. generalized the stem of the oblique cases (e.g. Gen. sg. *lig-os). For similar examples of generalization of voiceless stops see Matasovic 2010b. A different explanation is proposed by de Bernardo Stempel (1999: 42) who supposes that the ending –ach is due to analogy with adjectives witht he productive suffix –ach < *-āko-.

REF: de Bernardo Stempel 1999: 42. *lesko- ‘weak, lazy’ [Adj] GOID: OIr. lesc [o] gl. piger W: W llesg ‘weak, frail, faint, feeble’ ETYM: It is doubtful whether the GL PNs Liscus, Lisca belong here. The IE etymology is unclear. In principle, it is formally possible to derive PCelt. *lesko- from *lexsko- < *legh-sko-, from the root *legh- ‘lay, lie down’ (see *leg-o-) and compare OHG ir-lescan ‘extinguish’ (Germ. löschen), but a deverbal adjective with the present suffix *-sko- would be surprising. REF: GPC II: 2157, Elamarre 204. *lessu- ‘benefit, use’ [Noun] GOID: OIr. les [u m] ‘relief, advantage, good’ W: MW lles [m] ‘use, profit, advantage’; ‘gainful, profitable’ CO: OCo. les gl. commodum ETYM: In spite of several suggestions in the literature, none of the IE etymologies proposed so far appears convincing. A derivation from the root *pleh1- ‘be full’ is formally impossible. A compound from the same root is probably found in OIr. díles ‘own, loyal, steadfast’ and W dilys ‘pure, authentic, valid’ < *dī-lessu-, but the semantic connection is not obvious. REF: GPC II: 2156, Irslinger 2002: 111f. *lextu- ‘tomb’ [Noun] GOID: OIr. lecht [u m] ‘tomb, death’ W: MW lleith [m] ‘death, destruction’ (GPC llaith) PIE: *legh- ‘lie’ (IEW 658f.) COGN: Lat. lectus‘bed’ SEE: *leg(y)o- ‘bed’ ETYM: OIr. lecht is sometimes considered to be a borrowing from Lat. lectus, but the parallels in Welsh make it probable that the word is inherited. MW lleas, llias ‘death, bloodshed’ may represent a derivative from the same root (*legastu-). REF: GPC II, Stüber 1998: 158, Irslinger 2002: 110. *lisso- ‘walled space (about a dwelling), yard’ [Noun]

25

GOID: OIr. les [o m] ‘yard, palace, hall’ W: MW llys [m and f] BRET: MBret. les, MoBret. lez ‘yard’ CO: Co. lys ‘yard’ ETYM: These words are usually derived from the PIE root *pleth2- ‘broad’ (Gr. platýs, Lith. platùs, etc., IEW 833), in which case the correct reconstruction would be PCelt. *flisso-. However, I do not find this etymology convincing from the semantic point of view, so I tentatively reconstruct *lisso-. REF: GPC , Irslinger 2002: 283f., Stifter 1998: 215, Zair 2010: 238. *līwo- ‘accusation’ [Noun] SEE: *liy-o- ‘charge, accuse’ *loferno- ‘fox’ [Noun] W: MW llewyrn [Collective, sometimes m] ‘foxes’, W llwyrn, llwyrn ‘will-o’-the-wisp, ignis fatuus’ BRET: MBret. louarn, MoBret. louarn, lowern [m] CO: OCo. louuern gl. vulpes, Co. lowarn PIE: *h2lop- ‘fox’ COGN: Skt. lopāśá- ‘fox, jackal’, Gr. alō pēks ‘fox’, Lith. lãpė ‘fox’, Arm. ałuēs ‘fox’ ETYM: In OIr. we can compare the PN Loarn (parallel to OW Louern), perhaps also Gaul. PN Louernios. The parallels in other IE languages do not allow to reconstruct a PIE prototype precisely. This might be a Wanderwort of some kind. REF: GPC II: 2172, Deshayes 2003: 475, Delamarre 208, Schrijver 1998. *losko- ‘lame’ [Adj] GOID: OIr. losc [o] ‘lame, crippled’ PIE: *lokso- ‘oblique’ (IEW 308) COGN: Gr. loksós ‘slanting’ ETYM: If this etymology is correct, the OIr. word underwent a metathesis *lokso- > *losko-. The original form of the root might be preserved in W llechwedd [m] ‘hillside, slope, acclivity’, which may represent PCelt. *lexsowiyo-. In Gaul., one may compare the ethnonym Lexouii and (with the same metathesys as in Goidelic) PN Loscus. Of course, all of this is very speculative. REF: GPC II: 2126, Delamarre 201, 209. *lowano- ‘dirty’ [Adj] BRET: MBret. louan PIE: *lew- ‘dirty’ (IEW 681) COGN: Lat. pol-luo ‘make dirty, infect’ SEE: *lutā ‘mud, dirt’ ETYM: Although this adjective is attested only in Breton, it is both formally and semantically derivable from the root *lew-, which is also found in the zero-grade in PCelt. *lutā ‘mud, dirt’. The suffix *-ano- was generalized from roots where the adjectival suffix *-no- was preceded by a syllabic laryngeal. REF: Zair 2010: 286. *lustā ‘tail’ [Noun] GOID: MIr. los [o m and ā f] ‘end, tail, result’ W: MW llost [f] ‘tail, spear’ BRET: MoBret. lost ‘tail’ CO: MCo. lost ‘tail’ ETYM: I find improbable the derivation of these Celtic words from the PIE root *lewH- ‘cut off, loose’ (Gr. lýō, OIc. liósta ‘hit’, Go. fraliusan ‘lose’, LIV 417f.) REF: GPC II: , Schrijver 1995: 412, Irslinger 2002: 269. *lutā ‘mud, dirt’ [Noun] GOID: OIr. loth [ā f] ‘mud, mire’ GAUL: Luto-magus, Lutetia (Paris) [Toponyms] PIE: *lew-to- ‘mud, dirt’ (IEW: 681) COGN: Lat. lutum ‘mud’, po-lluo ‘make dirty, infect’, Alb. lum ‘slime, dirt’

26

ETYM: In W, lluddedic ‘muddy’ is derived from *lowd-eto- (dissimilated from *lowt-eto-) with the full grade of the same root, but this may be a ghost-word (lluddedig in GPC II: 2219 means ‘tired’. A Celtiberian cognate might be incorporated into lutiakos (A.76) ‘belonging to *Lutia’, where *Lutia would be a toponym derived from PCelt. *lutu-. Lith. liūtýnas ‘puddle’ is sometimes adduced as belonging to the same PIE root, but it is probably unrelated (the accentuation points to a laryngeal root). REF: GPC II: 2219, EIEC 160, Delamarre 211, Lewis 1989 42f., MLH V.1: 238, Irslinger 2002: 352, Zair 2010: 182. *magestu- ‘field’ [Noun] SEE: *magos ‘plain, field’ *mak-o- ‘increase, raise, feed’ [Vb] GOID: OIr. do-formaig ‘increase, add’; do-forma [Subj.]; do-formacht [Pret.] W: MW magu ‘feed, produce, rear’ BRET: MBret. maguaff ‘feed, rear’, MoBret. maezhur, magañ ‘feed, rear’ CO: Co. maga PIE: *meh2ḱ- ‘thin, elongated’ (IEW: 709 (*meh-)) COGN: Lat. maceo ‘be lean’, Gr. makrós ‘big’, Hitt. maklant- ‘thin, meagre, slim (of animals)’, OHG magar ‘meagre’ SEE: *makwo- ‘son’ ETYM: OIr. do-formaig < *to-wor-mak-o- (the simplex is unattested in Goidelic). The Celtic forms presuppose the zero-grade of the root, PIE *mh2ḱ- > PCelt. *mak-. The development of meaning would have been from ‘make thin’ to ‘make long’ and, finally ‘increase’ (cf. also Gr. makrós ‘great’ from ‘*long, elongated’). It is quite uncertain whether OIr. mér [o m] ‘digit, finger’ also belongs here; formally, it would be derivable from *mak-ro-, but the semantic development is difficult (from ‘thin/long finger’ to ‘finger’?). REF: KPV 466, GPC III: 2316, Deshayes 2003: 485, Zair 2010: 81. *mar-na- ‘betray’ [Vb] GOID: OIr. marnaid, -mairn (DIL mairnid) ‘betray, deceive, confound’; meraid, -mera [Subj.]; méraid, -méra [Fut.], mirt, -mert [Pret.]; mrathae, -mrath [Pret. Pass.] PIE: *merh2- ‘crumble, destroy’ (IEW: 735f.) COGN: Skt. mṛnā ti ‘crumble’, Gr. márnamai ‘fight’, ON merja ‘hit’ ETYM: The connection with the PIE root *merh2- is problematic on semantic grounds. The passive -mrath as well as the verbal noun mrath [o n] ‘betrayal, treason’ is unusual, since **mráth would be expected from *mrh2to-; the short vowel is probably the result of Dybo’s law. The comparison with MW brat [m and f] ‘betrayal, treason’, MCo. bras ‘id.’, OBret. brat gl. seditione shows that the participle *mrato- was substantivized in PCelt. REF: KPV 477ff., LIV 440, LEIA M-21, 67, Wagner 1967, McCone 1991: 15, 30, McCone 1996: 52f., Irslinger 271. *mati- ‘good’ [Adj] GOID: OIr. maith [i] W: MW mad BRET: MBret. mat, MoBret. mad CO: Co. mas GAUL: mat. (Coligny) PIE: *meh2-t- (IEW: 693) COGN: Lat. mānus ‘good’, Mātūta ‘goddess Dawn’ ETYM: OIr. math [u m] ‘bear’ is often understood as a tabooistic replacement for the original word for ‘bear’ (PCelt. *arto-) and related to this root. The u-stem (*matu-) may be attested in the Gaul. PN Matu-genos. Celtib. matus is sometimes also compared, but the meaning of this word (presumably a noun) is unknown (MLH V.1: 247f.), so it is not entitled to an etymology. Beyond Celtic, Gr. matís ‘great’ (Hesych.) is uncertain (it is not even ascertained that the word is Greek, it might be Galatian). The connection of these words to PIE *meh2- ‘give a sign, wink’ (Gr. mēnýō, Lith. móti, LIV 425) is very speculative. The development would have been from ‘wink’ to ‘give a sign of approval’ to ‘approve’, hence ‘good, approved’ as the meaning of the adjective from the same root. On the other

27

hand, if PCelt. *mati-, Lat. mānus and mātūrus ‘ripe, mature’ are related to Hitt. mehur ‘time’, the original meaning of PIE *meh2- was probably ‘timely’. Note, though, that Kloekhorst (2008: 568) derives Hitt. mehur from the root *meyh2- ‘diminish’ (Lat. minuo, etc.), in which case it cannot be related to PCelt. *mati-. REF: LEIA M-12f., GPC III: 2300, Deshayes 2003: 484, EIEC 235f., Lambert 1994: 109, Irslinger 2002: 150f., 207f., MLH V.1: 247f., Zair 2010: 80. *med-o- ‘measure, judge’ [Vb] GOID: OIr. midithir, -midethar; mestair, -mestar [Subj.]; miastair, -miastar [Fut.]; mídair [Pret.]; messae, -mess [Pret. Pass.] W: MW meddu ‘possess, rule’ CO: Co. medhes ‘say’ PIE: *med- ‘measure’ (IEW: 705f.) COGN: Lat. medeor ‘heal’, modus ‘measure’, Gr. médomai ‘care for, think of’, Go. mitan ‘measure’ ETYM: OIr. mess [u m] ‘judgement’ is derived from the same root (PCelt. *messu- < PIE *med-tu-), cf. also OIr. tomus ‘measure, mass’ < *to-messu-, commus ‘control, power’ < *kom-messu- and coimdiu [t m] ‘lord, master’ < PCelt. *kom-med-wot- (a substantivized perfect participle). For the possibility that this verb originally had athematic i-inflexion (Kortlandt 2007: 137) see *ar-yo- ‘plough’ REF: KPV 478ff., LIV 423, GPC III: 2394, LEIA M-48f., LP 381f., Irslinger 2002: 116f. *Medu-geno- [PN] SEE: *medu- ‘mead, alcoholic drink’ *mero- ‘crazy, silly’ [Adj] GOID: OIr. mer [o] ‘crazy, wild’ W: MW meredig, mereddig ‘foolish, strange’ ETYM: LEIA adduces as the Brittonic cognates MW mererid, meryerid ‘foolish, crazy’ which are not found in GPC (merierid ‘pearl’ is of course another word, borrowed from Lat. margarita). This could be due to simple confusion. In Gaulish, the element Mero-, -mero- in compound names (e.g. Es-merius) may be from the same root. It is assumed here that the formation found in MIr. is original, while Welsh shows a derivative from the same root. None of the deeper etymologies suggested by LEIA (s. v. mer) are more than guesses, but a connection with Gr. mōrós ‘stupid’ appears possible (see PCelt. *mar-o-). If this is correct, the Greek word would have to be derived from *mōrh2o-, with the lengthened grade of the root, while the Celtic forms would be from *merh2-o-. The development of meaning would be from ‘slow’ to ‘silly’ and ‘crazy’. REF: LEIA M-39, Delamarre 225, de Bernardo Stempel 1999: 42, Zair 2010: 176. *messu- ‘acorn’ [Noun] GOID: OIr. mess [u m] ‘acorns, tree-fruit’ W: MW mes [p] mesen [Singulative] BRET: MBret. mes, MoBret. mes [Collective] CO: Co. mesen ETYM: I find it difficult to accept the usual connection to PIE *med- ‘measure’ (IEW 705f.). Go. mats, OE mete ‘food’ would fit semantically, but these words are plausibly derived from *mh2d- ‘be wet’ (Gr. madáō, etc., cf. Orel 2003: 262); the primary meaning of PGerm. *matiz would have been ‘soft, wet food’ as opposed to ‘hard food’. However, PCelt. *messu- cannot be related to this root. It may have been borrowed from some non-IE language, although it is conceivable that it is derived from *met-tu-, where *met- is the root found in *met-o- ‘reap, cut’ (MW medi etc.). If so, the original meaning would have been ‘that which is cut, reaped’. REF: LEIA M-43, GPC III: 2438, Deshayes 2003: 506, Irslinger 2002: 117f., Wodtko et alii 2008: 463. *met-o- ‘reap, cut’ [Vb] W: MW medi BRET: MBret. midiff CO: Co. midzhi (MoCo.)

28

PIE: *met- ‘throw, reap’ (IEW: 703) COGN: Lat. meto ‘reap’, OCS mesti, metati ‘throw’, Lith. mèsti ‘throw’ SEE: *metelo- ‘reaper’ ETYM: LIV does not relate this PIE verbal root to OCS metati, Lith. mèsti ‘throw’, but I believe they belong here, rather than with a putative root *met- ‘reckon’ (Gr. métron, etc.), or with PIE *h2meh1- ‘mow’ (Gr. amáō, Eng. mow). The use of this particular verb as an agricultural term is one of the few exclusive Italic and Celtic lexical innovations. This root is probably contained as the second member of the compound OIr. deimes [o m] ‘scissors’ if it is from *dwi-met-to-. REF: KPV 483, GPC III: 2391, LIV 442, LP 54, Mallory & Adams 2006: 168, LEIA D- 43, Irslinger 2002: 251. *mēto-, *mēti- ‘fat, soft’ [Adj] GOID: OIr. méth [o and i] ‘plump, fat’ W: W mwyd [m] ‘a soaking, moistening’, mwydion [p] ‘tender parts of the body’; mwydyn [m] ‘pulp, soft inner part’ PIE: *mey(H)-t- (IEW: 711f.) COGN: Lat. mītis ‘mild, mellow, juicy’ ETYM: IEW (711f.) assumes that these words are related to words like OCS milъ ‘dear’, Lith. míelas ‘dear’, etc., in which case the Balto-Slavic intonation would imply that the root contained a laryngeal. Cf. also Skt. máyas ‘enjoyment’ which may go to *moyHos. The same root could account for OIr. mín ‘tender, soft’ < PCelt. *mīno- < *miHno-, and W mwyn ‘mild, tender, gentle’, MoBret. moan ‘tender, thin’, OCo. muin ‘thin’ < *mēnV- < *meyHnV-. See, however, PCelt. *moyni- ‘treasure, precious object’, where a different etymology is proposed. REF: LEIA M-44, GPC III: 2517, Deshayes 2003: 515, Irslinger 2002: 253f., Zair 2010: 152f. *mī, *me ‘I’ [Pron, Acc. s] GOID: OIr. mé, me-s(s)e W: MW mi BRET: MBret. me CO: Co. my, me GAUL: -mi (as verbal suffix only) PIE: *(h1)meh1, *(h1)me (IEW: 702) COGN: Lat. mē(d), Gr. emé, Skt. mām ETYM: In PCelt. two forms must be posited, *mī and *me. The form with the short *e was later regularly lengthened to yield OIr. (emphatic pronoun) mé. MBret. me also comes from *me, while MW mi must be from *mī. Already in PIE, this was the suppletive Acc. sg. to the 1st person singular personal pronoun *(h1)eh2om (Lat. ego etc.) which left no traces in Celtic. The initial *h1- in the oblique stem might be reconstructed only on the basis of prothetic e- in Greek. For the source of the alternation *(h1)me / *(h1)meh1 see PCelt. *nu. REF: LEIA 25f., GPC III: 2542, EIEC 454, Falileyev 112f., Lambert 1994: 66, Deshayes 2003: 498, Katz 1998a: 83f. *mlāto/i- ‘tender, soft’ [Adj] GOID: OIr. mláith [i] W: MW blawd [m] ‘flour’ BRET: OBret. blot, MoBret. bleud [m] ‘flour’, CO: Co. blot ‘flour’, MCo. bles PIE: *melh2- ‘grind’ (IEW: 716) COGN: Lat. molo, Lith. málti SEE: *mal-o- ‘grind’ ETYM: The development in Celtic was from ‘ground’ to ‘tender, soft’, and the PIE proto-form was presumably the past participle of the verb ‘to grind’, *mlh2to-, or the derived adjective *mlh2ti- (which may be more probable in the light of the fact that Dybo’s law did not operate; we would expect *mlato- from *mlh2tó-). The Brittonic forms could also be from PCelt. *blātu- ‘flower’ with the same semantic development as in English flour < Fr. fleur (de farine). REF: LEIA M-56, GPC I: 284, LP 54, DGVB 86, Deshayes 2003: 115, de Bernardo Stempel 1999: 293, Hamp 1973: 151f., Schrijver 1995: 179, McCone 1996: 52, Irslinger 2002: 209f.

29

*nanī- ‘grandmother’ [Noun] W: MW nain [f] PIE: *nana- ‘mommy’ [IEW 754f.] COGN: Skt. nanā ‘mommy’, Gr. nánnē ‘aunt’, Alb. nënë ‘mother’ ETYM: Clearly a nursery word, so the parallels in other IE languages may be accidental. REF: GPC III: 2550. *nanti- ‘fight, battle’ [Noun] GOID: MIr. néit [i m] PIE: *nent- ‘combat, fight’ (IEW: 755) COGN: Go. ana-nanþjan ‘dare’, ON nenna ‘to strive’, OHG nenden ‘to dare’, ToA nati ‘might, strength’ ETYM: According to DIL, Néit may have been the name of an old Irish war-god. Gaulish personal names with the element Nanti- (e.g. Nantio-rix) may also belong here, but in many cases it is difficult to separate them from the element *nantu-, *nanto- ‘valley, stream’. The PIE etymology is uncertain. I doubt whether the Tocharian forms are related to this root (cf. also ToB nete ‘might, strength’), since the *-n- should have been preserved. REF: LEIA N-7, EIEC 201, Delamarre 231f., Ellis Evans 1967: 236f., Hamp 1976a, de Bernardo Stempel 1999: 284., Lindeman 1999, Irslinger 2002: 225f. *nāwā ‘boat’ [Noun] GOID: OIr. nau [ā f] (DIL nó) W: MW noe [f and m] ‘large vase, bowl’ BRET: MBret. neau, nev, MoBret. new [f] ‘vase’ PIE: *neh2u- ‘boat’ (IEW: 755f.) COGN: Skt. náu-, Lat. nāuis, Gr. naũs, Arm. naw, OIc. nōr ETYM: Cf. Gallo-Lat. nausum ‘ship’ (Ausonius). Originally an u-stem in PIE, this noun became a feminine ā-stem in PCelt. (cf. OIr. Gen. sg. naue). The Brittonic forms represent derivatives from PCelt. nāwā (MW noe < *nāwyā, MBret. nev < *nāwī). REF: LEIA N-5, GPC III: 2592, EIEC 74, Deshayes 2003: 537, Delamarre 232, de Bernardo Stempel 1999: 27, 32, McCone 1994: 73. *naw(i)nyā ‘hunger’ [Noun] GOID: OIr. naunae, noíne [iā f] (DIL núna) ‘hunger, famine’ W: MW neuynn (GPC newyn) [m] ‘hunger, famine’ BRET: MBret. naffn, MoBret. naoun, naon [m] CO: OCo. naun gl. faimis, Co. nown SEE: *nāwito- ‘need’ ETYM: The original OIr. form, naunae, was changed to noíne under the influence of oíne ‘fast’ (from Lat. ieiunium). W newyn could be from *nawinV- (with *-i- under the influence of *nāwito- ‘need’), but other reflexes point to *nawnV-. The PIE root could be *neh2w- (or *newh2-, with laryngeal metathesis) ‘need’ (OCS nužda ‘need’, nuditi ‘compel’, OPr. nautin [Acc. sg.] ‘need’, Cz. nutiti ‘compel, force’, ORuss. nyti ‘grieve’, Go. nauþs ‘need’, IEW 756). REF: LEIA N-21, GPC III: 2580, Deshayes 2003: 532, Uhlich 1995a: 23ff., Zair 2010: 286f. *nawito- ‘need’ [Noun] GOID: OIr. neóit [i] ‘narrow, stingy’ W: MW neued (GPC neued, neuedd, neufedd) [m] SEE: *naw(i)nyā ‘hunger’ ETYM: The exact development of OIr. neóit is unclear to me. It may be unrelated to W neued, although the etymology is almost generally accepted. The PIE root could be *neh2w- ‘need’ (Lith. naudà ‘need’, Go. nauþs ‘need’, IEW 756) as in *nāw(i)nyā ‘hunger’. REF: LEIA N-10, GPC III: 2576. *nemo- ‘poison’ [Noun] GOID: OIr. neim [s n] PIE: *nem- ‘divide, take’, *nemos ‘that, which is taken’ (IEW: 763) COGN: Go. niman ‘take’, Gr. némō ‘divide, take’

30

ETYM: Vendryès rejects this etymology on semantic grounds, but it is not hopeless (cf. Germ. Gift ‘poison’ from the verbal root which is in geben ‘give’). If the s-stem in OIr. is original, the correct reconstruction is *nemos-, and the word is homophonous with *nemos- ‘heaven, sky’. REF: LEIA N-7, de Bernardo Stempel 1999: 142. *nemos- ‘heaven, sky’ [Noun] GOID: OIr. nem [s n] W: OW nem [f and m], MW nef BRET: OBret. nem, MBret. neff, MoBret. neñv [m] CO: OCo. nef gl. celum, MCo. nef PIE: *nebhos ‘cloud, cloudy sky’ (IEW: 315) COGN: Hitt. nēpis ‘sky’, Skt. nábhas- ‘cloud’, Gr. néphos, OCS nebo ‘sky’ ETYM: In Gaulish, possible cognates are PNs Nemesia, Nemessi [p]. The irregular *-m- of the Celtic forms is best explained as the result of assimilation (n...bh > n...m). Another possibility is that–m- is due to the analogical influence of *nemeto- ‘sacred place, sanctuary’. REF: LEIA N-8, GPC III: 2561, DGVB 265, EIEC 110, Falileyev 118, Campanile 1974: 82, Delamarre 234f., de Bernardo Stempel 1999: 140, 142, 148, Deshayes 2003: 534, Stüber 2002: 131, Wodtko et alii 2008: 499. *now-slo- ‘cry, shout’ [Noun] GOID: OIr. nuall [o n] PIE: *newH- ‘cry, roar’ (IEW: 767) COGN: Skt. návate, ToB nu- ‘roar’ ETYM: Another possible cognate is Lat. nūntium ‘message’, but that word has also been plausibly related to PIE *newo- ‘new’. Mayrhofer (s. v. NAV-) thinks this root may be onomatopoetic. MoIr. núar ‘wail, lament’ is probably from the same root (*now-ro-). REF: LEIA N-24, LIV 456, Zair 2010: 276. *obnu- ‘fear’ [Noun] GOID: OIr. omun [u m] W: MW ofuyn (GPC ofn) [m] BRET: MBret. oun, MoBret. ovn [m] CO: Co. own GAUL: Ex-obnus [PN] SEE: *exs-obno- ‘fearless’ ETYM: OIr. omun is also attested with the long first vowel, ómun > MIr. úamun. The source of the vacillation is unclear to me (short vowel by analogy with essamin ‘fearless’?). Cf. also the denominative verb W ofnaf ‘to fear’. The PIE etymology is unknown, but a connection with Croat. jéza ‘fear, terror’ appears possible. Cf. also Lith. éngti ‘to strangle’, ON ekki ‘convulsive sobbing’, OE inca ‘doubt, complaint’ (Orel 2003: 84). The Slavic forms presuppose Early PSl. *endzā or *indzā, and can be derived from PIE *h1engw-, with the nasal infix from the verbal base (present stem), cf. also Lith. éngti (the acute is regular because of Winter’s law). The PIE root would be *h1egw-. PCelt. *obno- or *obnu- can be from PIE *h1ogwno- or *h1ogwnu-. A different etymology is proposed by McCone (1992b), who derives the Celtic forms from *fowno- and relates them to Lat. pavor and OIr. úath (see under *fowtu- ‘fear’). REF: LEIA O-22, GPC III: 2631, Delamarre 170, Deshayes 2003: 470, Schrijver 1995: 353, de Bernardo Stempel 1999: 94, Matasovic 2010. *olīnā ‘elbow, angle’ [Noun] GOID: OIr. uilen [ā f] W: OW elinou [p], MW elin [f and m] BRET: MBret. elin, MoBret. elin [m] CO: Co. elyn, elin PIE: *h3elēn ‘elbow’ (IEW: 307) COGN: Lat. ulna, Gr. ōlénē, Go. aleina, perhaps ToB alyiye ‘palm’ ETYM: Cf. also Gaul. hydronym Olina (Sims-Williams 2006: 97, today’s Orne). W elin is from *olin by i-Umlaut. The PIE reconstruction is uncertain, since the long vowel in Gr. ōlénē is unexplained (cf. also Latv. uôlekts ‘elbow’ with the broken tone pointing to the root *HeH-l-).

31

REF: LEIA U-18, GPC I: 1206, EIEC 98, 176, Falileyev 53, Deshayes 2003: 233, Delamarre 241, Matasovic 2004: 107, Sims-Williams 2006: 97, Zair 2010: 27f. *omo- ‘raw’ [Adj] GOID: OIr. om [o] W: W of PIE: *HoHmo- ‘raw’ (IEW: 777f.) COGN: Gr. ōmós, Skt. āmá-, Khot. hāma-, Arm. hum ETYM: As the Gr. and Skt. words are oxytona, this could be an instance of pretonic shortening in Celtic (Dybo’s law). Note, however, that there are but a few instances of this law operating on syllables with proper PIE vowels (*e, *o), so in this particular case one is tempted to think that the second laryngeal was lost by dissimilation (*HoHmo- > *Homo- > PCelt. *omo-). The PIE reconstruction is uncertain, since the nature of the second laryngeal cannot be ascertained (initial *h2 is usually reconstructed on the basis of Arm. hum). A connection with Lat. amārus ‘bitter’ is doubtful (it may, in principle, represent *h2H-emo-). Gaul. personal names with the element Omo- (e.g. Omos) might also be from this root, but this is far from certain. REF: LEIA O-21, GPC III: 2629, Delamarre 241, EIEC 478, Wodtko et alii 2008: 202, Zair 2010: 177. *ono- ‘blemish’ [Noun] GOID: OIr. on [o n] PIE: *h2enh3o- ‘blame’ (IEW: 779) COGN: Gr. ónomai ‘blame’, Hitt. hannari ‘litigates, sues’ ETYM: The etymology is speculative, since the OIr. form is short and there are no parallels in other Celtic languages. If it is correct, OIr. on must be from PIE *h2onh3o-, and the Greek vocalism must be due to assimilation *ano- > *ono- (as assumed by LIV 282). LIV reconstructs the root as *h2neh3-, but this would not work for Celtic *ono-. SEE: *anamī ‘blemish’ REF: LEIA O-22f., Zair 2010: 54f. *oxtū ‘eight’ [Num] GOID: OIr. ocht [Nasalizing] W: OW oith, MW wyth BRET: OBret. eith, MBret. eiz, MoBret. eizh CO: Co. eath GAUL: oxtumetos ‘eighth’ PIE: *h3eḱtoh1 ‘eight’ (IEW: 775) COGN: Skt. astáu, Lat. octō, Gr. oktō, Go. ahtau, Lith. aštuonì, Alb. tetë, ToB okt ETYM: Gaul. oxtumetos, OIr. ochtmad, and W wythfed, MoBret. eizfed point to PCelt. ordinal *oxtumeto- ‘eighth’ (cf. Skt. astamá-). REF: LEIA O-7, GPC III: 3746, EIEC 100, 402., Falileyev 124f., 153, LHEB 405, Delamarre 246, Deshayes 2003: 212, Schrijver 1995: 350. *oxtumeto- ‘eighth’ [Num] SEE: *oxtū ‘eight’ *oyto- ‘oath’ [Noun] GOID: MIr. oeth [o m] W: OW an-utonau gl. periuria [p] PIE: *h1oy-to- ‘a going’ (IEW: 295) COGN: OHG eid ‘oath’, ToB aittanka ‘directed toward’. ETYM: OW an-ut-on-au contains the negative prefix an- adnd a nasal suffix. In Gaul., we find PNs such as Oitoccius, Ar-oitus, which may be related. Presumably, PCelt. *oyto- is a derivative of the root *h1ey- ‘to go’; the original meaning of ‘oath’ in Celtic and Germanic would be ‘a going, path’. This could be motivated by the custom of walking between pieces of sacrificed animals to give force to the oath, cf. also Gr. oĩtos ‘course, fate’. REF: LEIA O-12, EIEC 408f., Delamarre 240, Falileyev 8, Benveniste 1969, II, ch. 3, Irslinger 2002: 260.

32

*razd-o- ‘scrape, scratch’ [Vb] W: MW rathu (GPC rhathu) BRET: MBret. razaff, MoBret. razhan, rahein (Vannetais) ‘scrape, shave’ PIE: *(H)reh2s-dh(h1)o- ‘shear, scratch, shave’ (IEW: 854 (*rēd-)) COGN: Lat. rādo ‘shave’ ETYM: A connection of these words with Germ. Ratte ‘rat’, Eng. rat, and the related Germanic words is possible, if their original meaning was ‘rodent’ vel sim. REF: KPV 528f., LIV 496, GPC III: 3040, Zair 2010: 56. *rēd-o- ‘ride’ [Vb] GOID: OIr. réidid, -réid; reraid [Pret.] PIE: *(H)reydh- ‘ride’ (IEW: 861) COGN: OE rīdan, Lith. riedė ti ‘roll’ SEE: *ufo-rēdo- ‘horse’, *rēdi-‘simple, easy’ ETYM: Also from this root are W rhwyd-hau ‘hurry’ and the Gaul. PN Ande-redus and ethnonym Redones (> Rennes, Redon). The same root is preserved in PCelt. *ufo-rēdo- ‘horse’ and, probably, *rēdi- ‘simple, easy’. REF: KPV 535, LEIA R-26, Delamarre 255. *rendi- ‘point, peak’ [Noun] GOID: OIr. rind [i m] BRET: OBret. gabl-rinn ‘fourche à pointe, forked branch’ ETYM: The OBret. compound gablrinn has the exact parallel in Ir. gabhal-rind. It is unclear whether OIr. rind [i n] ‘star’ is originally the same word, or an accidental homonym without etymology. The semantic connection would be clear if stars were mythologically represented as spear-points. A convincing IE etymology is lacking. It is unclear whether OIr. rét ‘star’ is related to rét ‘thing’ (see PCelt. *rentu-) or not, but the similarity with *rendi- ‘star’ is worth noting. REF: LEIA R-232, DGVB 297. *rentu- ‘thing, matter’ [Noun] GOID: OIr. rét [u m] PIE: *(H)ren-t- ‘possession’ (IEW: 865) COGN: Skt. rátna- ‘possession, precious object’ ETYM: Quite a speculative etymology, since the reflexes of the PIE root are limited to just two branches, and a metathesis of *nt > tn must be assumed in Sanskrit. Mayrhofer (s. v. rátna-) leaves the Sanskrit word without a PIE etymology. The derivation from PIE *h1rem- ‘be still’ (e.g. Skt. rámate ‘becomes still’, Lith. ram tis ‘support’, IEW 864, LIV 252) is not better, since the meanings are far apart. REF: LEIA R-22, de Bernardo Stempel 1999: 93, Irslinger 2002: 152f. *rinno- ‘quick, fast’ [Adj] GOID: MIr. renn [o] ‘quick, swift’, substantivized rend, renn ‘foot, leg’ W: MW rin (GPC rhyn) ‘unyielding, stiff, brave, fierce, cold’ CO: Co. rynny ‘trembling’ PIE: *h3reyH- ‘flow’ COGN: Skt. rinā ti ‘makes flow’, Gr. orī nō ‘whirl’, OHG rinnan ‘run’, OCS rějati ‘flow’ ETYM: The range of meanings of the reflexes of the assumed PCelt. *rinno- is considerable, so it is possible that two different etyma became confused. The origin of the geminate *-nn- is unknown. The usual derivation from *rinwo- is impossible, since *-nw- is preserved as OIr. -nb-, cf. OIr. menb < *menwo- ‘small’. It seems probable, however, that the PIE root is *h3reyH- ‘flow’. A derivative from this root *h3riH-n-do- would have given *rīndo- and then, by Osthoff-like shortening, *rindo- > *rinno-. The meanings ‘to flow’ and ‘to run, flee’ are often expressed by a single verb, cf. OCS tešti and Croat. tèći ‘flow, run’. REF: LEIA R-20, GPC III: 3140, LIV 305f. *rowtro- ‘assault, onrush’ [Noun]

33

GOID: MIr. rúathar [o m, perhaps originally n] W: MW ruthyr (GPC rhuthr) [m] ‘rush, attack, invasion’ PIE: *h3rew- ‘hurry, rush’ COGN: Lat. ruo, Gr. oroúō ETYM: LIV (510) reconstructs the root as *rewH- ‘tear up’ because of reflexes such as OIc. rýja ‘tear off wool’ and OCS ryti ‘dig’, but these may be from a different root. REF: LEIA R-49, GPC III: 3106, Zair 2010: 278. *sād-ī- ‘set,fix’ [Vb] GOID: OIr. sáidid, -sáidi ‘thrusts, fixes, implants’ W: MW gwahawd ‘invites’ (GPC gwahodd) SEE: *sed-o- ‘sit’ ETYM: MW gwahawd is from the prefixed *ufo-sād-. This Celtic verb is originally the causative to *sed-o- ‘sit’; the ō-grade is found in OCS saditi ‘set’, which is also from PIE *sōd-eye-. REF: LEIA S-8f., GPC II: 1562, LIV 513, Schumacher 2000: 161f. *sālo- ‘sea, ocean’ [Noun] SEE: *salano- ‘salt’ *sati-, *satyo- ‘swarm, throng’ [Noun] GOID: OIr. saithe [io m] W: MW heyd (GPC haid) [f]; heidiau [p] ‘a swarm (of bees), flock, pack (of hounds)’ BRET: MBret. het, MoBret. hed [m] CO: Co. hês PIE: *sh1-ti- (IEW: 890) SEE: *sīlo- ‘seed’, *satV- ‘seed’ ETYM: OIr. saithe is from PCelt. *satyo-, but the Brittonic forms point to an i-stem, *sati-, which may be primary. Possible Gaulish cognates include the PNs Sati-genus, Satia, Sattius, etc. The PIE root is *seh1- ‘sow’ (Lat. sero, etc.). A different etymology derives these words from PIE *seh2- ‘have one’s fill’, cf. Gr. áatos ‘insatiable’, Germ. satt ‘full, satiated’, Lat. satis ‘enough’ (IEW 876, see also Watkins 1979 and PCelt. *sāti-). If it is correct, the meaning ‘swarm’ must be derived through a metaphor ‘satiety of bees’ for which Watkins adduces some evidence. REF: LEIA S-16, GPC II: 1813, Deshayes 2003: 324, Delamarre 268, de Bernardo Stempel 1999: 445, 527, Watkins 1979: 191-194. *satV- ‘seed, offspring’[Noun] W: MW had [Collective] CO: MCo. has [Collective] BRET: MBret. hat, MoBret. had [m] PIE: *sh1-tV- (IEW: 890) SEE: *sīlo- ‘seed’, *sati-, *satyo- ‘swarm, throng’ ETYM: The vocalism of the suffix cannot be established. Since the attested forms are collectives, PCelt. *satā is a plausible reconstruction. REF: GPC II: 1799 , Zair 2010: 72. *sekw-o- ‘say’ [Vb] GOID: MIr. sechid, sichid; sich [Pret.] (DIL sichid) ‘asserts, declares’ W: OW hepp [3 s Pres.], MW hebu BRET: OBret. hep PIE: *sekw- ‘say’ (IEW: 895) COGN: Gr. enn-épō ‘tell’, Lat. in-seque ‘say!’, inquit ‘says’, OHG sagen SEE: *sekw- ‘follow’, *skwetlo- ‘story’ ETYM: W. ateb ‘answer’ is from PCelt. *ati-sekwo-, cf. also OIr. aithesc [o n] ‘answer’. The exact relationship of PIE *sekw- ‘say’ and *sekw- ‘follow’ is unclear, but there was probably a single verb with both meanings. REF: LEIA S-64f., GPC II: 1830, DGVB 209, LIV 526f. KPV 565f., Falileyev 83, Hamp 1974. *sferā ‘heel’ [Noun]

34

GOID: MIr. seir [ā f] W: MW ffer [f and m] ‘ankle’ (GPC ffêr) BRET: MBret. Fergant [PN] CO: OCo. fer gl. crus PIE: *sperh1-o- / *sprh1-o- ‘heel’ (IEW: 993) COGN: OE spor ‘footprint’, ToB sprāne [Dual] ‘flanks’ ETYM: MoBret. fer ‘heel’ is a loanword from W. (Deshayes 2003: 232). The MIr. dative sg. serid (Cormac’s Glossary) points to a dental stem (PCelt. *sferet- ?), and traces of a n-stem are probably preserved in W dial. uffarn ‘heel’, OCo. ufern gl. talus, MoBret. ufern, uvern, which may be from *ofi-sfer-no-. Gr. sphyrón ‘ankle(bone)’ is probably related, but the aspirate is unexpected. The vowel y in Gr. sphyrón ‘heel’ might be the result of Cowgill’s law (o > y between labials and resonants). The reflexes in other IE languages probably justify the reconstruction of a paradigm *sperh1- / *sporh1-m / *sprh1-os. Celtic generalized the full grade (*sperh1-), while Germanic generalized the zero-grade (PIE *sprh1o- > PGerm. *spura- > OE spor). Lubotsky (2006) relates these forms to Skt. sphuráti ‘kicks away’, Lat. sperno ‘push away, despise’ and reconstructs the root *tsperH- ‘kick with the heel’. The same root might, perhaps, be posited for Skt. pā rṣṇi- ‘heel’, Gr. ptérna ‘heel’, Lat. perna ‘leg, haunch’ (< *tperH-sneh2). REF: LEIA S-73, GPC I: 1283, EIEC 265, Deshayes 2003: 232, 755, Schrijver 1995: 348, 374, Campanile 1974: 46, Hamp 1982a, de Bernardo Stempel 1999: 172, Irslinger 2002: 59, Lubotsky 2006, Zair 2010: 299. *sflissi- ‘shaving, splinter’ [Noun] GOID: MIr. slis [i f] PIE: *spleyd- ‘split’ (IEW: 986) COGN: MHG splitter, Eng. split ETYM: MIr. slis is not well attested, but there is a more common synonymous derivative, slisiu [n f] which may be from PCelt. *slissyon-. The PCelt. form *sflissi- is from the zero-grade of the root (*splid-ti-), which is otherwise attested only in Germanic. REF: LEIA S-136, Stüber 1998: 137, Irslinger 2002: 217. *simV- ‘chain’ [Noun] GOID: MIr. sim ‘chain or loop used in securing a cattle pound’ P IE: *sh2ey- ‘bind’ (IEW: 892) COGN: OE sīma ‘rope, tie’, Hitt. ishāi ‘binds’, Skt. sinā ti ‘fetters’ ETYM: A very tentative etymology, since the MIr. word is attested only twice, and its gender and word-formation are unknown. For semantic reasons it can be plausibly connected to OE sīma ‘rope, tie’, but it is unclear why the –i- is short (maybe we should read sím, or the vowel was shortened by Dybo’s law). REF: LEIA S-110, Zair 2010: 159. *skwetlo- ‘story’ [Noun] GOID: OIr. scél [o n] W: MW chwedl [f and m] (GPC chwedl, chweddl) BRET: MBret. que-hezl, MoBret. kehel [m] ‘information, intention’ CO: Co. whethel PIE: *sekw- ‘say’ (IEW: 898) COGN: Lat. in-quam, OHG saga ‘tale’, Lith. sakýti ETYM: It is difficult to derive the Brittonic forms directly from PCelt. *skwetlo- (the expected outcome would probably have been initial *sp- > W ysp-, cf. *skwiyat- ‘hawthorn’). They may be loanwords from Goidelic, or we would have to assume an early metathesis of *skw- to *kws- > *xs-, with the subsequent regular development to W chwedl (cf. *xswib-ī- ‘move, recede’ > W chwythu). MoBret. kehel is from *kom-skwetlo-. The same root is probably found in MIr. scoth [ā f] ‘word’, which is found mostly in glossaries an poetic texts. It may be from PCelt. *skwotā. REF: LEIA S-39f., GPC I: 846, Deshayes 2003: 379, de Bernardo Stempel 1999: 302, Schrijver 1995: 375, Irslinger 2002: 63. *skāto- ‘shadow’ [Noun]

35

GOID: OIr. scáth [u and o n] W: MW ysgawd [m] ‘shade, darkness’ (GPC ysgod) BRET: MBret. squeut CO: OCo. scod gl. umbra, MCo. schus ‘terror’ PIE: *skeh3t- ‘darkness, shadow’ (IEW: 957) COGN: OE sceadu, Gr. skótos SEE: *ufo-skāto/u- ‘shelter, shadow’ ETYM: The Celtic forms are sometimes derived from PIE *skōto-, which would represent a vṛddhi-formation from the root *skot- ‘shadow’, while the Germanic and Greek words are derivable from *skotu- and *skoto- respectively. It is far more probable that we should assume a PIE root-noun with the ablaut pattern *skeh3t- / *skh3t-os, with Celtic preserving the full grade of the root, and Germanic and Greek the zero-grade from the oblique cases. REF: LEIA S-36, GPC III: 3838, EIEC 508, de Bernardo Stempel 1999: 324, 528, Irslinger 2002: 125ff., Zair 2010: 140f. *skerd-(y)o- ‘scrape off’ [Vb] GOID: OIr. scerdaid, -sceird ‘peels, scrapes off’; scaird [Pret.] (DIL sceirtid) PIE: *skerdh- ‘cut, pierce’ (IEW: 940) COGN: Lith. skerdžiù ‘I cut’ ETYM: A derivative from the same root is probably found in MIr. scris [u m] ‘scraping, tearing’ < *skrissu- < *skrdh-tu. In all likelihood the PIE root *skerdh- is ultimately the same as *skerH- ‘cut, pierce’ (cf. PCelt. *skara-). REF: KPV 582, LIV 558, LEIA S-39, Irslinger 2002: 127. *skīto- ‘tired’ [Adj] GOID: OIr. scíth [o] ‘weary, tired’ W: MW escud, esgud (GPC esgud) ‘quick, vivid’ BRET: MBret. escuit ‘quick’ CO: Co. uskys ‘quick, vivid’ PIE: *skeh1t(H)- ‘injure, harm’ (IEW: 950) COGN: Go. skaþjan ‘damage’, OHG scado ‘harm, loss’ ETYM: The Brit. forms may be from a compound *exs-skīto-, where *exs- is a negative prefix (cf. PCelt. *exsobno- ‘fearless’). The vocalism of MW esgud (for expected *esgid) is unaccounted for, cf. OIr. escid ‘tireless’, and scís [o and u m], also [ā f] ‘fatigue, tiredness’ < *skīssu- < *skeh1d-tu-. Gr. a-skēthē s ‘unharmed’ might be related, but th (instead of *t) in this word is difficult to account for. OHG scado and the related Germanic forms must be from *skh1t(H)-, with the zero-grade of the root. A different etymology is proposed by Schrijver (1992), who thinks that the Brittonic words are borrowed from Goidelic and reconstructs PCelt. *skwīto- (perhaps from the same root as in Lat. quiēs ‘sleep, rest’ < PIE *kwyeh1- (OCS po-čiti ‘to rest’), IEW 638). REF: LEIA S-46f., GPC I: 1245, EIEC 312, Schrijver 1992: 8f., Irslinger 2002: 300f., Zair 2010: 238. *(s)kītto- ‘left, clumsy’ [Adj] GOID: MIr. cittach [o] ‘left-handed, awkward’ W: MW chwith ‘left, left-handed, sinister, sad, wrong’ PIE: *skh2ey- ‘left’ COGN: Lat. scaeuus, Gr. skaiós ETYM: MIr. has also the variant cettach, showing a-affection, expected if the i was short. The form cittach might point to *kīt-, but i is never written long. W chw- is from the initial cluster *sk- metathesised to *ks-. The Greek and Latin words for ‘left’ can be derived from *skh2ey-wo-, while in order to account for the Celtic forms we must assume the zero-grade (*(s)kh2it-) and ‘expressive’ gemination (*(s)kh2i-tto-), which renders this etymology rather speculative. A different etymology of MW chwith is proposed by Schrijver (2003), who derives the Welsh word from PIE *ksweybh- ‘make a swift movement’ (LIV 373), from which we have PCelt. *xswib-ī- ‘move, recede’. However, this is difficult to reconcile with MIr. cittach. REF: LEIA C-108, EIEC 349, GPC I: 858, Schrijver 2003. *skoltā ‘fissure, cleft’ [Noun] GOID: MIr. scoilt, scolt [ā f] ‘splitting’

36

W: MW hollt [m and f] BRET: MBret. faout [m] ‘fissure’ CO: Co. fals ‘fissure’ PIE: *skel(H)- ‘divide’ (IEW: 924) COGN: Hitt. iskallai- ‘tear apart, slit open’, Gr. skállō ‘hoe’, ON skilja ‘divide’, Lith. skélti ‘split’ ETYM: It is difficult to derive the British and Irish forms from exactly the same PCelt. prototype, but they are definitely related, see LEIA S-48f. We would expect a prothetic y- in Welsh (*skoltā > *ysgollt), so W hollt must be derived by metathesis (PCelt. *sko- > *xso- > *xwo- > ho-), and MoBret. faout is even more difficult to account for, perhaps by *xw- > f-, which is not a well established sound change. Deshayes (2003: 228) denies the connection, and derives faout from OFr. falder ‘fold, wrinkle’, but the semantic connection is weak. In Irish, scoilt is attested late, only in Modern Irish. However, there is a denominative verb scoiltid ‘break, split’ in OIr., cf. also the W denominative verb hollti, holli ‘slit, split, open’. In PIE *skel(H)- the laryngeal is assumed because of the acute in Lith. skélti (it could have been lost in Celtic regularly after *ol by Saussure’s rule). REF: LEIA S-48f., GPC II: 1894. Deshayes 2003: 228, Schrijver 1992, Irslinger 2002: 358, Zair 2010: 311f. *slig-o- ‘strike, hew’ [Vb] GOID: OIr. sligid, -slig; sléiss, -slé [Subj.]; silis, -sil [Fut.]; selaig [Pret.]; slechtae, -slecht [Pret. Pass.] PIE: *sley- ‘smear, creep’ (IEW: 663f.) COGN: OHG slīhhan ‘sneak’, OCS slьzъkъ ‘slippery’, perhaps Hitt. salik- ‘to touch, have contact with’ ETYM: The meanings ‘smear’, ‘sneak’ and ‘hit’ are not easily reconciled, but cf. Germ. Streich ‘blow’ and streichen ‘smear’ (LIV 566). It is possible that OIr. slicht [u m] ‘sign, trace, version’ contains the same root (PCelt. *slix-tu- < *sli-tu-), but the semantic connection is weak. W llith ‘bait’, which is sometimes compared with OIr. slicht, is probably unrelated. REF: KPV 591f., LIV 566f., LP 396, LEIA S-133, Irslinger 2002: 127f. *slimno- ‘polished, smooth’ [Adj] GOID: OIr. slemon,slemun, slemain [o] ‘smooth, sleek, polished’ W: OW limnint [3p Pres.] ‘polish’, MW llyfn ‘polished, smooth’ BRET: OBret. limn gl. lentum, MBret. di-leffn ‘hard’, MoBret. levn ‘smooth’ CO: Co. leven PIE: *sley-m- ‘smear (with grease), polish’ COGN: Lat. līmo ‘polish’, OHG slīmen ‘polish’ ETYM: The oldest form of the OIr. word appears to be slemun, with the same development of the final syllable as in domun ‘world’ < *dumno- < *dubno-. The PCelt. word is built from the root *sleym- with the suffix adjectival suffix *-no-. REF: LEIA S-130, GPC II: 2254, DGVB 242f., EIEC 527, Falileyev 103, de Bernardo Stempel 1999: 437, 540, Zair 2010: 185. *slu-n-k-o- ‘swallow’ [Vb] GOID: OIr. slucaid, -slucai; sloicis [Pret.]; -slogeth [Pret. Pass.] (DIL sluicid, slocaid) W: MW llyncu BRET: OBret. ro-luncas gl. guturicauit, MBret. louncaff PIE: *(s)lewk-, *(s)lewg- (IEW: 964) COGN: Gr. lýzō ‘swallow’, Ukr. lýkaty ‘swallow’, Germ. schlucken ‘swallow’ ETYM: The root-form *(s)lewk- is a variant of *(s)lewg- (IEW 964), from which we have Germ. schlucken, Gr. lýzō ‘to have the hiccup’, etc. and OIr. loingid ‘eat’ (< PCelt. *lu-n-g-o-), W llewa ‘eating, drinking’ < *lug-ā-. The variation *k/g is unexplained, but such irregularities are not uncommon in expressive verbs. It is also possible that root final *-g is original, while *-k was generalized from the sigmatic aorist, where *-gs- yielded *-ks- regularly. REF: LEIA S-139, GPC II: 2273, KPV 593f., LIV 568, LEIA S-139, LP 396, Stüber 1988: 66. *snīs ‘we’ [Pron] GOID: OIr. sní

37

W: MW ni BRET: MBret. ni, ny CO: Co. ny PIE: *no- ‘we’ (IEW: 758) COGN: Lat. nōs, Hitt. anzas ‘us’, Go. uns ‘us’, OPr. noūson, Alb. ne SEE: *swīs ‘you’ ETYM: Gaul. sní in line 3 of the Chamalières inscription probably represents the same pronoun, but this text is written in scriptio continua, and the segmentation of words is uncertain. The vocalism of PCelt. *snīs may have been influenced by *swīs ‘you’, and the analogy with this pronoun might also account for the initial *s-. The infixed form of the pronoun (OIr. -nn-, MW -n-) may reflect the original *snōs (LP 215). Katz (1998a) reconstructs the PCelt. form as *snī (without final *-s) and derives it from PIE *ns-me, which supposedly developed as *nsne (dissimilation) > *ansne > *sne (aphaeresis) > *snī (lengthening in monosyllables). I find this complex derivation difficult to believe. REF: LEIA 150f., GPC III: 2581, LP 215, Katz 1998a: 54f., 100f. *soybo- ‘crooked’ [Adj] SEE: *sēbro- ‘demon, spectre’ *stlondo- ‘sense, meaning’ [Noun] GOID: OIr. slond [o ?m] ‘expression, mention, designation’ W: OW istlinnit [3s Pres.] gl. profatur, W ystlwn ‘connection’ BRET: MBret. stlen ‘declared’ ETYM: Cf. also the OIr. verb sluindid ‘expresses, declares, names’ from the same root (causative *stlond-ī-). Schrijver (1995: 435ff.) defends the old connection of these words with Lat. splendor ‘brightness’, Lat. splendeo ‘shine’, cf. also ToB pällātär ‘praise’, Go. spill ‘tale’. If this is true, the correct PCelt. reconstruction is *sflondo-, and the semantic development would have been from ‘enlighten, shed light on’ to ‘declare’. Cf. also OIr. lés ‘daylight, rays’ which could represent *sflanssu- < *splnd-tu- from the same root. REF: LEIA S-137f., Falileyev 97, Schrijver 1995: 435ff., de Bernardo Stempel 1999: 295, Irslinger 2002: 111. *sukko- ‘pig’ [Noun] GOID: OIr. soc(c) [o m] ‘snout, plough-share’ W: OW huch, suh, MW hwch [f and m] ‘pig’ BRET: OBret. hoch gl. aper, MoBret. houc’h CO: OCo. hoch gl. porcus, Co. hoch PIE: *suH- ‘pig’ (IEW: 1038) COGN: Lat. sūs, Av. hū-, ToB suwo, OE sugu ‘sow’, Alb. thi ETYM: Cf. also the OIr. river-name Socc and socc sáil ‘name of a fish’. Fr. soc ‘plough-share’ is from Gaul. *sukko- and shows the same development of meaning as OIr. socc. Cf. also the Gaul. PNs Succus, Sucio, Succius, etc. The geminate *-kk- may be due to the analogical influence of the other word for ‘pig’, PCelt. *mokku-. The short vowel may be due to the generalization of the root shape in the oblique cases of the root noun (*suHs, Gen. sg. *suH-os). The velar suffix and the short vowel are also found in Lat. sucula ‘small pig’ and OE sugu ‘sow’. REF: LEIA S-158f., GPC II: 1928, EIEC 425, LHEB 567, Falileyev 87, Campanile 1974: 64, Gamillscheg 811, Delamarre 285, de Bernardo Stempel 1999: 35, 96, 519, Zimmer 1994: 152-3, Testen 1999, Zair 2010: 206. *swan-na- ‘sound, play (an instrument)’ [Vb] GOID: OIr. seinnid, -seinn; sifais, -sif(e) [Subj.]; sefainn, sephainn [Pret.] PIE: *swenh2- ‘(produce) sound’ (IEW: 1046f.) COGN: Skt. svan-, ásvanīt [3s Aor.], Lat. sono, sonus ‘sound’, OE swinn ‘music’ ETYM: PCelt. *swan-na- is from the zero-grade of the PIE root (*swn-neh2-), with the regular development of syllabic *n > *an before resonant. MIr. séis [i f] ‘melody, sound, music’ may be from PCelt. *swen-sti-, with the generalized shape of the root without laryngeal. REF: KPV 607f., LIV 611, LP 395, LEIA S-86, S-75, Watkins 1963: 130-133, Irslinger 2002: 424f. *swarrinā ‘gland, ulcer’ [Noun]

38

SEE: *swerwo- ‘bitter’ *swekru- ‘mother-in-law’ [Noun] W: MW chwegr [f] CO: OCo. hweger gl. socrus PIE: *sweḱru-h2 ‘mother-in-law’ (IEW: 1043f.) COGN: Skt. śvaśrū-, Lat. socrus, OCS svekry, Arm. skesur, Alb. vjehërr, OE sweger SEE: *swekrVno- ‘father-in-law’ ETYM: The PIE form of the word for ‘mother-in-law’, *sweḱruh2, should have been reflected as *swekrū in PCelt., but the absence of i-affection in Welsh shows that the vowel of the final syllable was shortened (as in Latin). It is also possible that the proto-form was *swekrū-, but that the Welsh form is chwegr rather than *chwygr on the analogy with chwegrwn ‘father-in-law’. REF: GPC I: 848, LP 18, Campanile 66, Zair 2010: 301. *swensti- ‘melody, music’ [Noun] SEE: *swan-na- ‘sound, play (an instrument)’ *swerwo- ‘bitter’ [Adj] GOID: OIr. serb [o] W: MW chwerw BRET: MBret. hueru, MoBret. c’houero CO: Co. wherow SEE: *sworo- ‘louse’ ETYM: W chwarren [f] ‘gland, ulcer, lump’, if related, must represent PCelt. *swarrinā < *swarsinā (with *ar < syllabic *ṛ). These Celtic words lack a persuasive IE etymology. A connection to PIE *swer- ‘wound’ (IEW 1050) is possible (Av. xvara- ‘wound’, OHG swero ‘pain’); for the semantic development cf. the connection between Eng. bitter (PGerm. *bitraz) and bite (PGerm. *bītanan). Nikolaev 2010 compares Luv. šihuwa/i- ‘bitter’, which may be from PIE *seh2/3- ‘sharpness’ (cf. also the Germanic words from ‘sword’, e.g. Germ. Schwert). The Celtic forms would be derivable from an adjectival *sHwer-wo-. REF: LEIA S-90, GPC I: 345, 850, Morris Jones 1913: 29, Nikolaev 2010: 54. *swīs ‘you (p)’ [Pron] GOID: OIr. sí W: OW hui, MW chwi BRET: OBret. hui, MBret. huy CO: Co. why GAUL: (?) sui, sue (Châteaubleau tile); suis (Chamalières) ETYM: Both Gaul. forms sui < *swīs and sue < *swes are found in a single inscription, the Châteaubleau tile (lines 2-5). The interpretation of suis from Chamalières is uncertain, like the interpretation of most other forms in that text (see Lambert 1994: 158). OIr. hapax emphatic pronoun sib (Wb. 19c20), as well as its reflex in MoIr. sibh, may be from reduplicated *swe-swe, which would imply the existence of a form with a short vowel (*swe, parallel to *me ‘I’ besides *mī). The shape of the PCelt. 2 pl. personal pronoun is unusual; the PIE stem is *we-, attested mostly in ō-grade (Lat. uōs, OCS vy, etc.), but the word-initial *s- must be analogical (perhaps after the reflexive pronoun *swe-). Katz’s (1998a: 100ff.) derivation of PCelt. *swī (without final *-s) from *swe (with lengthening in monosyllables) < PIE *uswe (by aphaeresis) is improbable, in my view. OIr. sethar ‘your (pl.)’, proclitic far n-, is from *swes-tero- (cf. Lat. uos-ter, ues-ter). REF: LEIA S-101f., GPC I: 850, GOI 254, 282, Falileyev 87f., Lambert 1998-2000, Katz 1998a: 87f. *tanxto- ‘frozen, thick’ [Adj] SEE: *tanko- ‘peace’ *tāti-/*tā-nt- ‘thief’ [Noun] GOID: OIr. táid [nt, m] gl. fur PIE: *teh2-y- ‘steal’ (IEW: 1010) COGN: Skt. tāyú- ‘thief’, OCS tatь ‘thief’, tajiti ‘hide’, Hitt. tāyezzi ‘steals’

39

ETYM: The word formation of OIr. táid is unclear. It is an nt-stem, i.e. a nominalized participle of the PIE verb PIE *teh2-, which is unattested in Cletic, but the Nom. sg. táid cannot be directly inherited from PIE, since word-final *-ant-s would be reflected as *-e in OIr. (cf. *karant- ‘friend’ > OIr. carae). Rather, it may be compared to OCS tatь, which is a masculine i-stem (presumably an old abstract ‘the stealing’). REF: LEIA T-7, LIV 616, EIEC 543, de Bernardo Stempel 1999: 285, 355ff., Irslinger 2002: 213f. *tawso- ‘deaf’ [Adj] GOID: OIr. tó [o], túae [yo] W: MW taw [m] ‘silence’ BRET: MBret. tao, MoBret. taw [m] ‘silence’ CO: Co. taw SEE: *taws-yo- ‘be quiet, be still’ ETYM: Derived from the verbal root *taws- < PIE *teh2us- (IEW 1056f., for cognates see *taws-yo-). Taul. PN Tausius may also be from the same root. REF: LIV 642f., GPC IV: 3455, LEIA 90f., Delamarre 2003: 293, Deshayes 2003: 716., Zair 2010: 90. *tegu- ‘thick’ [Adj] GOID: MIr. tiug [u] ‘thick, dense, solid’ W: OW teu, MW tew ‘thick, fat’ BRET: MBret. teo, teu CO: Co. tew PIE: *tegu- ‘thick’ (IEW: 1057) COGN: OHG dicchi, OE ðicce ‘thick’ ETYM: While MIr. tiug is clearly an u-stem, the Brittonic forms go back to (secondary) thematized *tegwo-. Reflexes of this PIE root are attested only in Celtic and Germanic. Both branches point to an original u-stem (Orel 2003: 411). REF: LEIA T-76, GPC IV: 3491, Falileyev 147, Schrijver 1995: 68, Uhlich 1993, Malzahn 2011. *teterV- ‘scaldcrow’ [Noun] GOID: MIr. tethra PIE: *teter- ‘a kind of bird’ (IEW: 1079) COGN: Skt. tittirá- ‘partridge’, Lith. tetervà ‘capercaille’, Gr. tetráōn ‘capercaille’ ETYM: MIr. tethra is attested mostly in glossaries. Its stem-formation cannot be ascertained. In some contexts it refers to the war-goddess (Badb). Delamarre (295) compares the Gaul. NP Tetarus, Tettaro, but this may be just a chance correspondence. The PIE word, whatever its precise meaning, was obviously onomatopoetic. REF: LEIA T-56, Delamarre 295, Mallory & Adams 2006: 143f. *tixtā ‘going, coming’ [Noun] GOID: OIr. techt [ā f] W: MW teith (GPC taith) [f] ‘voyage, trip’ SEE: *tēg- ‘go’ ETYM: In OIr., techt is the verbal noun to téit ‘goes’. The Celtic forms adduced here can be derived from PIE *stigh-teh2. Derivatives from the same root include MW tith [f] ‘gallop, swift movement’ (< *stixti-) and MW tuth [m] ‘trot’ < (*stoyxto-). REF: LEIA T-60, de Bernardo Stempel 1999: 52, Schumacher 2000: 69. *tlāti- ‘weak, sick’ [Adj] GOID: MIr. tláith [i] ‘weak, feeble, soft, pleasant’ W: MW tlawdd ‘poor, sick’ PIE: *telh2- ‘support’ (IEW: 1060f.) COGN: Gr. talás ‘sad’, OHG dolen ‘suffer’ SEE: *talamon- ‘earth’, *tli-na- ‘take away, steal’ ETYM: Cf. also MIr. tlás [o m and ā f] ‘weakness, mildness’ < *tlāstā. The meaning in Celtic can be derived from ‘suppressed’, and finally from the PIE root *telh2- ‘support’, but complex semantic evolution must be assumed (‘supporting’ > ‘suppressed’ > ‘weak’, or ‘support’ > ‘suffer’ ‘be weak,

40

sick’). The form *tlāti- can be both from PIE *tl h2-ti- (because Dybo’s law did not operate), or from PIE *tleh2-ti- (where both the initial and the final accent would be possible). REF: LEIA T-78, GPC III: 3510, Hamp 1985: 184, Schrijver 1995: 187, McCone 1996: 52f., de Bernardo Stempel 1999: 293, 337, Irslinger 2002: 214. *togo- ‘roof’ [Noun] SEE: *tegos ‘house’ *trātu- ‘length of time, hour’ [Noun] GOID: OIr. tráth [u n] W: MW trawd [m and f] ‘course, voyage’ (GPC trawd, trawdd) PIE: *terh2- ‘traverse, cross’ (IEW: 1074) COGN: Skt. tárati ‘crosses’, OCS trajati ‘to last’ SEE: *taras- ‘across’ ETYM: PCelt. *trātu- presupposes PIE *tŕh2tu- or *tréh2tu-. The development of meaning was from ‘traversing, passing’ to ‘duration, period’ and ‘time’. REF: LEIA T-124f., GPC IV: 3560, de Bernardo Stempel 1999: 287, Schrijver 1995: 188f., Irslinger 2002: 135f., Zair 2010: 98. *trātV- ‘weak’ [Noun] W: MW trawd BRET: MoBret. treut COGN: Lat. tero ‘to rub’, Gr. teírō ‘to oppress, distress, weaken’, téretron ‘drill’, Russ. terét’ ‘to rub’. PIE: *terh1- ‘rub, drill’ (IEW: 1071f.) ETYM: PCelt. *trātV- presupposes PIE *tŕh1tV-, but the same root is attested with the suffix (‘enlargement’) *-y- in OIr. tríath [o] ‘weak’ < *trēto- < *trey(h1)to- and OIr. tréith [i] ‘weak’ < *trēti- < *trey(h1)ti-. For the suffix cf. Lat. trītum ‘rubbed’, trītor ‘who rubs or grinds’ < *trih1-tōr. The development of meaning in Celtic was from ‘rubbed, oppressed’ to ‘weak’. REF: LEIA T-134., GPC IV: , LIV 632, Irslinger 2002: 214f., Zair 2010: 99, 280. *trexs(n)o- ‘strong’ [Adj] GOID: OIr. trén [o], Ogam TRENA-GUSU W: MW trech ‘stronger’ BRET: MBret. trech, MoBret. trec’h [m] ‘victory’ CO: Co. trygh ‘victorious’ GAUL: Trexius, Trexa, Trenus [PN] PIE: *treg- ‘strength’ (IEW: 1090) COGN: OE þraka ‘courage’, ON þrekr ‘strength’ ETYM: The Brittonic and Gaulish forms are from underived *trexso-, whereas OIr. trén is from *trexsno-. OIr. tress [u m] ‘contention, fight’ is from *trexsu-, while W tres [m and f] ‘raid, battle, attack’ must be a borrowing from Goidelic (*xs yields ch in Welsh). Certain cognates are attested only in Germanic and point to PGerm. *þrakja- < PIE *trogyo- (Russ. trógat’ ‘touch’ adduced by IEW is unrelated). REF: LEIA T-136, GPC IV: 3571, 3587, Delamarre 301, Deshayes 2003: 739, de Bernardo Stempel 1999: 252, 258, 389, Ziegler 1994: 117. *trexsu- ‘fight’ [Noun] SEE: *trexs(n)o- ‘strong’ *trīkont- ‘thirty’ [Num] GOID: OIr. trícho [nt m] BRET: OBret. tricont, trigont, MBret. tregont GAUL: tricontis SEE: *trīs ‘three’ ETYM: In MW there is a (possible) hapax trychwn ‘thirty’. The word for ‘thirty’ is a compound of *trī- ‘three’ and the element –kont- for which see *kwenkwekont- ‘fifty’. REF: LEIA T-140, GPC IV: 3637, Delamarre 301, Zair 2010: 304.

41

*tuxtu- ‘form’ [Noun] GOID: OIr. tucht [u m] PIE: *twek- ‘be visible’ COGN: Hitt. dukkāri ‘is visible’, Skt. tvác- ‘skin’ ETYM: The derivation from PIE *tuk-tu- is straightforward and the semantic connection is convincing, but the verbal root *twek- is attested only in Anatolian. The connection with Gr. teúkhō ‘make’ (PIE *(s)tewgh- ) is less probable. REF: LEIA T- 166, LIV 654, Irslinger 2002: 136. *ufo-ber(w)o- ‘source, well’ [Noun] GOID: MIr. fobar (DIL fofor) [ā f], MoIr. fobhar [m] GAUL: Uobera [Hydronym] (> Vavre, Vaivre), [Toponym] (> Vabre, Voivre) SEE: *berw-ā-, *berw-ī- ‘boil, cook’, *beruro- ‘watercress’, *brutu- ‘fermentation, boiling’ ETYM: MIr. fobar is poorly attested, but MoIr. fobhar is quite a common word. These words are very likely derived from the same root as in *berwā- ‘boil, cook’, but the absence of evidence for root-final *-w- is puzzling. REF: Delamarre 325, DGVB 145. *ufo-kliyo- ‘north’ [Noun] GOID: OIr. fochla [yo n] W: gogledd [m] BRET: MoBret. gwalez ‘northern wind’ SEE: *kliyo- ‘left’ ETYM: The motivation of the development from ‘left’ to ‘north’ lies in the fact that the north is on the left side when one is facing the east (the rising sun). REF: GPC II: 1436. *ufo-lawto- ‘property, wealth’ [Noun] GOID: OIr. folud [o n] ‘substance, material, property, equivalent, reason, cause’ W: MW golud [m and f] CO: OCo. wuludoc gl. dives PIE: *leh2u- ‘benefit, prize’ (IEW: 655) COGN: Gr. apolaúnō ‘enjoy’, Lat. lucrum ‘profit’, OE lēan ‘reward’, Go. laun ‘reward’ ETYM: The Celtic forms represent a derivative with the suffix *-to-, or the past participle (*leh2u-to-). Cf. also OIr. lúag, lóg ‘prize’ < *lowgo- < *loh2u-go-. OE lēan and Go. laun can be from *leh2w-no-. Lat. lucrum may represent *lu-tlo-, but the loss of laryngeal is unusual (by Dybo’s rule?). OIr. lour ‘enough’ and W llawer ‘many, a lot’ (MCo. lower), llawen ‘merry’ (MBret. louen, Co. lowen) could also be from the same root (*law-ero-, *law-eno- ?, cf. Gr. lārós ‘tasty’ < *lāwaros), and OIr. deolaid [i] ‘bestowed by favour, gratuitous’ is from *dī-ufo-lawti-. REF: GPC II: 1452, 2109, EIEC 484, de Bernardo Stempel 1999: 144, Delamarre 198, Irslinger 2002: 256f., Zair 2010: 274. *wastu- ‘dwelling’ [Noun] W: MW gwas [f] ‘abode, dwelling’ PIE: *weh2stu- ‘dwelling’ (IEW: 1170f.) COGN: Skt. vā stu- ‘house, dwelling’, Gr. ásty ‘city’, ToB ost ‘house’ ETYM: OIr. i foss ‘at rest, at home’ is sometimes related to this set of forms, but the o-vocalism makes this comparison unlikely. It is more probably related to foss ‘rest’, which can, in turn, be derived from *ufo-sto- (see PCelt. *wasto- ‘servant’). MW gwas may reflect the zero-grade of the PIE word, which had an ablauting paradigm (presumably *woh2stu- / *wh2stew-). For the development *wHC- > *waC- in Celtic, see *waxto- ‘bad’ (the same development is found in Gr. ásty < *wh2stu-). A different etymology derives both OIr. foss and MW gwas from PIE *h2wos-to- (with the change *wo- > W gwa- as in *wolto- ‘hair’ > W gwallt), from the PIE root *h2wes- ‘spend the night, abide’ (Hitt. huiszi ‘lives’, Go. wisan ‘be, remain’, Arm. goy ‘is’, IEW 72, 1170f., LIV 293). REF: GPC II: 1591, EIEC 281, Irslinger 2002: 257.

42

*waydā ‘shout, cry, clamour’ [Noun] GOID: OIr. fáed, faíd W: MW gwaedd [f] ETYM: Possibly connected with *way ‘woe’. Theoretically, one could think of a compound *way-dhh1-eh2, with the root *dheh1- ‘do, make’ as the second element (Lat. facio, Gr. títhēmi, OCS děti, etc., IEW 235ff.), but this is quite far-fetched, since no trace of such a compound is attested in any other IE language. REF: GPC II: 1548. *waylino- ‘sea-gull’ [Noun] GOID: OIr. faílenn, foílenn [o m] W gwylan [f] BRET: MoBret. gouelan CO: OCo. guilan gl. alcedo ETYM: The Brittonic forms probably represent the feminine *waylanā parallel to the masculine *waylino- in Goidelic. A derivation from the interjection *way is formally possible, but semantically not quite convincing (see also *waylo- ‘wolf’). The word may be a borrowing from some non-IE substratum in Insular Celtic. REF: GPC II: 1760, Schrijver 1995: 115f., Matasovic 2011. *weni-kar-o- [PN] SEE: *wenyā ‘family’ *wenkyo- ‘crossbeam’ [Noun] GOID: MIr. féice [yo m] ‘ridge-pole’ PIE: *wenḱyo- (IEW 1112) COGN: Skt. vámśya- ‘cane, bamboo-cane, upper beams’, Wakhi was ‘roof-beam’ ETYM: A very tentative etymology, as it relies on the correspondence of the MIr. word and a Sanskrit word attested only in Atharva-Veda and later. A connection with the verb *wenk-ā- ‘bend’ is possible for MIr. féice, but not for the Sanskrit word, which has the reflex of the palatalized velar, while *wenk-ā- is from PIE *wenk- ‘bend’ with a plain velar. REF: Zair 2010: 87. *wer-V- ‘find’ [Vb] GOID: OIr. -fúar [Pret.]; fo-fríth ‘was found’ [Pass. Pret.] PIE: *werh1- ‘find’ (IEW: 1160) COGN: Gr. heurískō, Arm. gerem ‘take prisoner’ ETYM: OIr. -fúar represents a reflex of the PIE reduplicated perfect (*we-worh1-). The same root is sound in OIr. fríth [o m], fríthe [yo m] ‘that which is found, foundling, waif’, which is from PCelt. *wrīto- < *wreh1-to-, cf. also Gaul. Ateuritus, Uritius [PN]. REF: EIEC 202, LIV 698, Irslinger 2002: 252, Delamarre 329. *wextā ‘time, course, turn’ [Noun] GOID: OIr. fecht [ā f] W: MW gweith (GPC gwaith) [f] ‘1. time, course, 2. work, act’ BRET: OBret. gueid ‘time’, MBret. guez, MoBret. gwezh [f] CO: OCo. gueid gl. opus, MCo. gweth, gwyth ‘time’ PIE: *weh- ‘carry, drive’ (IEW: 1118ff.) COGN: Lat. ueho ‘carry’, Gr. ókhos ‘cart’, OHG wegan ‘move’, Lith. vèžti, OCS vezetъ ‘carries’ SEE: *wegno- ‘wagon’ ETYM: If this etymology is correct, the PCelt. word is the past participle of the PIE verb ‘to carry, convey’ (Lat. uectus); the semantic development was from ‘carried away’ to ‘passed, gone’, to ‘period of time’. This is possible, but not compelling. OIr. fecht is also used to denote an expedition (‘usually with hostile intentions’ according to DIL); in this latter sense it may be parallel to OW guyth, MW gwyth [m] ‘rage, fury’, from PCelt. *wixtā / *wixto- < PIE *weyḱ- ‘fight’ (cf. PCelt. *wik-o-). Gaulish and Brittonic ethnonyms in –vices (e.g. Ordo-vices, Lemo-vices) probably also contain the same root, and it is also found in personal names such as Gaul. Veco-rix, which may be parallel to OIr. Fíachrai (Gen. sg. Fíachrach).

43

REF: GPC II: 1563, 1790, Deshayes 2003: 309, Delamarre 309, 318, Irslinger 2002: 251f. *winto- ‘wind, whistling’ [Noun] GOID: OIr. fet [ā f] ‘a whistling or a hissing sound’ W: MW gwynt, gwint [m] ‘wind’ BRET: MBret. guent, MoBret. gwent [m] ‘wind’ CO: OCo. guins gl. ventus, Co. gwyns PIE: *h2weh1-nto- ‘wind’ (IEW: 81ff.) COGN: Lat. uentus, Skt. vā ta-, Hitt. huwant-, OHG wint, ToB yente ETYM: If the etymology of OIr. fet is correct, we have to assume Osthoff-like shortening in Celtic (*wēnto- > *wīnto- > *winto-), since PCelt. *wīnto- would yield OIr. *fít. On the other hand, OIr. feth [m] ‘breeze’ (not féth, as adduced in DIL) may be from *wito-, with unexpected short –i- (by Dybo’s law?). REF: GPC II: 1778, Deshayes 2003: 305, McCone 1996: 55, 63, 107, Zair 2010: 209. *wissu- ‘knowledge’ [Noun] GOID: OIr. fius [u and o m] W: MW gwys [m and f] ‘declaration, summons’ PIE: *weyd- ‘see’ (IEW 1125ff.) COGN: Skt. vindáti ‘finds’, Lat. uideo, Arm. gitem ‘know’ SEE: *wēd-o- ‘find’ ETYM: The Celtic forms represent an abstract tu-stem (*wid-tu-) from the root *weyd- ‘to see’. There are many derivatives in OIr. and MIr., e.g. cubus ‘conscience’ < *kom-wissu-, ros ‘great knowledge’ < *fro-wissu-, nós ‘custom’ < *nowo-wissu- ‘new knowledge’, etc. REF: GPC, LIV 665ff., Irslinger 2002: 100ff. *wītyo- ‘woven, plaited’ [Adj] GOID: OIr. fíthe [yo] COGN: Lat. uieo ‘plait, weave’, Skt. vyáyati ‘bends, encloses’, Lith. výti ‘twist, wind’ SEE: *wi-na- ‘bend, wrap’ REF: Zair 2010: 152. *wiwero- ‘squirrel’ [Noun] GOID: OIr. íaru [n f], Scottish Gaelic feòrag W: W gwywer BRET: MoBret. gwiber PIE: *wi-wer- ‘squirrel’ (IEW: 1166) COGN: Lat. uīuerra ‘ferret’, OPr. weware ‘squirrel’, OE āc-weorna ‘squirrel’, Lith. vaiverìs ‘male polecat or marten’, Cz. veverka ETYM: OIr. íaru probably generalized the lenited variant of the word, hence the absence of initial f-. This word is certainly related to the Brittonic forms, although the formal correspondence is not perfect. REF: GPC , EIEC 540, de Vaan 2008: *wox-tlo- ‘dispute’ [Noun] GOID: OIr. focull, focal [o n, later m] ‘dispute’ W: MW gwaethl [m] ‘dispute, debate’ PIE: *wokw- ‘voice, word’ (IEW: 1135f.) COGN: Skt. vā k-, Lat. uōx, OHG giwahanem ‘recall’, Arm. gočem ‘call’ SEE: *wekwo- ‘face’ ETYM: OIr. focal, focul [o n] is often assumed to be a Latin loanword (cf. Lat. uocābulum), but the development *xtl- > *-kl- > -cul is attested in OIr. anacul < *anextlo-. The same root (*wokw-) appears in OIr. an-ocht ‘a metrical fault’ < PCelt. *an-uxto- < *n-ukwto-, parallel to Skt. anukta-, with the zero-grade of the root. MIr. fúaimm [n n, later f] ‘sound’ may be from *wōxsman-, or rather from the prefixed *ufo-woxsman-, and the same prefix would account for MIr. fúach [o m] ‘word’ < *ufo-wokwo-. W gwep [f] ‘face, grimace’ is often also derived from PCelt. *wekwo-, with the e-grade, and compared to Gaul. PNs such as Uepo, Uepo-litanos ‘broad-faced’ (?), as well as MoBret. goap ‘joke’

44

but here the semantic evolution would have to be very complex if this etymology is correct (‘word’ > ‘joke, mocking’ > ‘grimace’ > ‘face’). REF: GPC II: 1552, EIEC 534f., Delamarre 314, Stüber 1998: 70f., de Bernardo Stempel 1999: 299. *wrastā ‘shower’ [Noun] GOID: OIr. frass[ā f] ‘rain-shower, torrent’ PIE: *h1wers- ‘shower, rain’ (IEW: 78ff.) COGN: Hitt. warsa- ‘fog, mist’, Skt. vars- ‘rain’, varsā- ‘rainy season’, Gr. hérsē ‘dew’ (Hom. eérsē) ETYM: It is not completely certain that OIr. frass (also spelled fros) is a feminine ā-stem, but most attested forms are compatible with this interpretation. PCelt. *wrastā can be the result of a regular ‘liquid metathesis’ between a labial and a dental (*warstā > *wrastā), in which case it preserves the zero-grade of the PIE root (*h1wrs-to-). REF: Hamp 1981a, Irslinger 2002: 346f., Zair 2010: 34. *wrīto- ‘found, that which is found’ [Noun and Adj] SEE: *werV- ‘find’ *yalo- ‘clearing’ [Noun] W: MW ial (GPC iâl) GAUL: Eburo-ialum (> Ebreuil), Verno-ialum (Verneuil) [Toponyms], perhaps Ialonus [Theonym] ETYM: Cf. also MW an-ial ‘wasteland’. The existence of the simplex MW ial has been doubted. Pokorny’s (IEW: 504f.) comparison with PSl. *jalъ ‘waste, bare, unfruitful’ (Russ. jálovyj, jályj, Cz. jalový) is possible under the assumption that PIE *yHC- would be reflected as PCelt. *yal- rather than *īl-. For the possibility that *wHC- likewise gives PCelt. *waC- rather than *ūC- cf. PCelt. *waxto- ‘bad’. REF: GPC II: 1999, Lambert 1994: 39, Delamarre 186. *yālo- ‘praise, worship’ [Noun] GOID: OIr. áil, ál ‘request, act of asking’ W: MW iawl [?f] ‘prayer, supplication, worship’ BRET: OBret. iolent gl precentur PIE: *(H)yeh2lo- ‘zeal’ (IEW: 501) COGN: Gr. zē los ‘zeal’, Croat. jâl ‘envy’ ETYM: OIr. áil is usually used with the copula in phrases such as is áil do... ‘it is desirable to...’, cf. also the denominative verb áilid ‘wishes’. Cf. also OBret. iolent [3 pl.] gl. precentur, W iolaf ‘praise’. REF: LEIA A-30f., GPC II: 2002f., Zair 2010: 68. *yextV- ‘speech, language’ [Noun] GOID: MoIr. icht [?u m] ‘people, tribe’ W: MW ieith [f] ‘language, nation, race’ (GPC iaith) BRET: MBret. yez, MoBret. yezh [f] ‘language’ CO: Co. yēth PIE: *yek- ‘say, speak’ (IEW: 503f.) COGN: OHG jehan ‘speak’, Lat. iocus ‘joke’ ETYM: The stem and gender of Early MoIr icht are uncertain (it is a rare word). If it was a feminine i-stem, then the PCelt. reconstruction *yexti- would be probable. Moreover, if it were an u-stem, we would not expect raising of *e to *i (cf. OIr. recht ‘outburst of anger’ < *rextu-). However, a pre-form *yexti- would be reflected as *(i)th in Welsh, so the Brittonic do not seem to correspond exactly to MoIr. icht. A possible Gaulish cognate of these Insular Celtic forms is found in the inscription on the Châteaubleau tile (Lambert 1998-2000), where on line 2 we read iexsetesi, which might mean ‘you will say’ (McCone 2006: 101). This might be the 2 pl. future of a denominative verb from the root *yext-. Skt. yā cati ‘ask, solicite, entreat’ is unrelated (LIV 311, s. v. *yek-). REF: GPC II: 1999, LIV 311, Deshayes 2003: 761, Schrijver 1995: 106f., Irslinger 2002: 205f. *yowdo- ‘lord’ [Noun] W: OW Iud-, MW ud [m] BRET: OBret. Iud-cant [PN] PIE: *(H)yudh- ‘fight’

45

COGN: Lat. iūbeo ‘order’, Skt. yúdhyati ‘fights’ ETYM: Although the attested forms in Celtic are short, the etymology is formally and semantically irreproachable. The loss of *y- in MW ud in stressed syllable is regular. REF: GPC IV: 3699, Schrijver 1995: 280, Zair 2010: 68f.